HISTORY OF ARTIFICIAL INTELLIGENCE IN MEDICINE

Since the earliest stage of modern AI research, substantial efforts have been made in the medical domain. A script-based chatbot named ELIZA was proposed in 1966 [20]. ELIZA’s most famous script, DOCTOR, could interact with humans as a Rogerian psychotherapist. A biomedical expert system, MYCIN, presented in 1977, could analyze infectious symptoms to derive causal bacteria and drug treatment recommendations [21]. Later, in 1992, the probabilistic reasoning-equipped PATHFINDER expert system was developed for hematopathology diagnosis, to deal with uncertain biomedical knowledge efficiently [22,23].

Before the era of DL, several ML methods have been used widely in the medical domain. Moreover, the invention of digital medical imaging such as digital X-ray imaging, computed tomography and magnetic resonance imaging enabled computerized image analysis, where AI achieved another success in the medical domain. In 1994, Vyborny and Giger [24] reviewed the efforts to use ML algorithms featuring computer vision in several mammography analysis tasks, including microcalcification detection, breast mass detection and differentiation of benign from malignant lesions. They demonstrated the efficacy of computer-aided detection (CAD) by comparing the performance of radiologists with CAD to that of radiologists only. Later, in 2001, Kononenko [25] overviewed the typical ML methods such as decision trees, Bayesian classifiers, neural networks, and k nearest neighbor (k-NN) search, then reviewed their use in medical diagnosis and proposed evaluation criteria including performance, transparency, explainability and data resiliency. In 2003, however, Baker et al. [26] pointed out that the performance of commercial CAD systems was still below the expectation (max case sensitivity 49%) in detecting architectural distortion of breast mammography.

After the success of deep CNN in image classification, a wide range of attempts were made to apply DL to medicine. A notable success was the work of Gulshan et al. [27] in 2016, where retinal fundus images were analyzed by a CNN-based DL model to detect diabetic retinopathy lesions, achieving an area under the receiver operating characteristic curve (AUC) of 0.991, sensitivity of 97.5% and specificity of 93.4% in the high sensitivity setting, measured on the EyePACS-1 data set. In 2017, Litjens et al. [28] reviewed major DL methods suitable for medical image analysis and summarized more than 300 contributions in the neuro, retinal, pulmonary, breast, cardiac, abdominal, and musculoskeletal areas as well as in the digital pathology domain; contributions were well categorized according to their inherent type of image analysis: classification, detection, segmentation, registration, etc. Kohli et al. [29] presented another review on the application of ML to radiology research and practice, where transfer learning and data augmentation were emphasized as a viable solution to datalimited situations. Shaikhina and Khovanova [30] proposed another solution for a similar situation; their proposed solution incorporates multiple runs and the surrogate data test, which exploits statistical tools to guide the trained ML model having better model parameters and not being overfitted to a small training data set.

Genomics and molecular biology have been strongly connected to the medical domain since genome sequencing became real. Next-generation sequencing (NGS) technology allows a whole genome sequence to be translated into text composed of ATCG, so that necessary computational analysis can be done for disease diagnosis and therapeutic decision making. In 2016, Angermueller et al. [31] reviewed DL methods and their application to genomic and biological problems such as molecular trait prediction, mutation effect prediction, and cellular image analysis. They thoroughly reviewed the whole process used to apply DL to their problems, from data acquisition and preparation to overfit avoidance and hyperparameter optimization. Torkamani et al. [32] presented a review of high-definition medicine, which is applied to personalized healthcare by using several kinds of big data, including DNA sequences, physiological and environmental monitoring data, behavioral tracking data and advanced imaging data. Surely, DL techniques can help in analyzing those big data datasets in parallel, to provide exact diagnosis and personalized treatment. Another review was done in 2018 by Wainberg et al. [33] on the use of DL in various biomedical domains, including quantitative structureactivity relationship modeling for drug discovery and identification of pathogenic variants in genome sequences. They re-emphasized the importance of the performance, transparency, model interpretability and explainability of DL methods, in earning the trust of stakeholders gaining adoption. Besides these reviews, there exist two notable contributions for genetic variants. Xiong et al. [34] presented a computational model for gene splicing, which can predict the ratio of transcripts with the central exon spliced in, within the whole set of transcripts spliced from any given sequence containing an exon triplet. Recently an award-winning deep CNN-based variant caller named DeepVariant was announced [35], which is able to call genetic variation in aligned NGS read data by learning on images created upon the read pileups around putative variant sites.

Another type of medical data to be analyzed is electronic health records (EHR). Rajkomar et al. [36] recently published their work building a DL model that predicts multiple medical events, including in-hospital mortality, unplanned readmission, and prolonged length of stay, entirely from raw EHR records based on the Fast Healthcare Interoperability Resources format. Their model could accurately predict mortality events, with an AUC of 0.90 at patients’ admission, and even with an AUC of 0.87 at 24 hours before admission to the hospital. EHR data can be used in the prediction of other types of events, e.g., outcome of a patient biopsy, which could be predicted with AUC 0.69 in the work of Fernandes et al. [37]

Besides the analytical diagnostic tasks, AI has been tried in other areas, for example, an intelligent assistant named Secretary-Mimicking Artificial Intelligence that helps in the execution of a pathology workflow was presented by Ye [38]. Treatment decision is another important factor in patient healthcare, from both prognostic and financial perspectives. Markov decision analysis is an effective tool in such situations, which was used to solve the cardiological decision problem in the work presented by Beck et al. [39] Schaefer et al. [40] reviewed the medical treatment modeling using the Markov decision process, which is a modeling tool that fits well in the optimization of sequential decision making and is strongly related to reinforcement learning [41].

ARTIFICIAL INTELLIGENCE APPLICATION IN PATHOLOGY

Microscopic morphology remains the gold standard in diagnostic pathology, but the main limitation to morphologic diagnosis is diagnostic variability in bearing error among pathologists. The Gleason grading system is one of the most important prognostic factors in prostate cancer. However, significant interobserver variability has been reported when pathologists have used the Gleason grading system [42,43]. In order to get a consistent and possibly more accurate diagnosis, it is natural to introduce algorithmic intelligence in the pathology domain, at least in the morphological analysis of tissues and cells. With the help of digital pathology equipment varying from microscopic cameras to whole slide imaging scanners, morphology-based automated pathologic diagnosis has become a reality. In this review, we focus on morphology-based pathology: diagnosis and prognosis based on the qualitative and quantitative assessment of pathology images. Typical digital image analysis tasks in diagnostic pathology involve segmentation, detection, and classification, as well as quantification and grading [44]. We briefly introduce typical techniques used for AI in digital pathology and a few notable research studies per disease. The list of studies reviewed in this paper is given in Table 2.

TYPICAL TECHNIQUES

Digital pathology images used in AI are mostly scanned from H&E stained slides. Pathology specimens undergo multiple processes, including formalin fixation, grossing, paraffin embedding, tissue sectioning, and staining. Each step of the process and the different devices and software used with the digital imaging scanners can affect aspects of the quality of the digital images, such as color, brightness, contrast, and scale. For the best results, it is strongly recommended to alleviate the effect of these variations before using the images in automated analysis work [45]. Normalization is one of the techniques used to reduce such variations. Simple linear range normalization based on the equation [v_new = (v_old-a)/f_scale + b] is generally used for the pixel values in grayscale images, or for each channel of color images [47,60]. Scale normalization has not been reported in related works, as they all have used a single image acquisition device, e.g., a certain microscopic camera or digital slide scanner. When multiple image acquisition devices are used, scale normalization is of concern, since images acquired from different devices can have different pixel sizes, even at the same magnification level.

Detecting the region-of-interest (ROI) has been done by combining several computer vision operations, such as color space conversion, image blurring, sharpening, edge detection, morphological transformation, pixel value quantization, clustering, and thresholding [67]. Color space conversion is often done before pixel clustering or quantization, to separate chromatic information and intensity information [53]. Another type of color space conversion targets direct separation of color channels for hematoxylin (H), eosin (E) and diaminobenzidine from stained tissue images to effectively obtain nuclei area [57,59,66,68]. Thresholding based on a certain fixed value leads to low-quality results when there are variations in luminance in the input images. Adaptive thresholding methods like hysteresis thresholding and Otsu’s method can generate better thresholding results [47,53,59,69]. Recently, pixel-wise or patch-wise classifiers based on CNN have been used widely in ROI detection [44,49-51,54-56,58,65], where a deep CNN is trained to classify the type of target pixel or patch centered on the larger input image patch in a sliding window manner. Semantic segmentation CNN is another recent trend for this task [65,70,71], which can detect multiple ROIs in a given image without sliding window operation, resulting in much faster speed.

In the development of a CNN-based automated image analysis, data-limited situations are common in the medical domain, because it is very costly and time-consuming to build a large amount of annotated, high-quality data [45]. As previously mentioned, transfer learning and data augmentation should be incorporated to get a better result. In transfer learning, convolutional layer parameters of a CNN, pre-trained with a well-known dataset like ImageNet, are imported into the target CNN as layer initialization, while later layers like fully connected layers or deconvolutional layers are initialized randomly [62,70,71]. Additional training steps can update all of the layer parameters, including the imported ones, or only the parameters of the layers that were randomly initialized. With sufficient data, building a model without transfer learning is reported to give better performance [54].

A common strategy of image data augmentation is, for the given image, applying various transformations that do not alter the essential characteristics; such transformations include rotation (90°, 180°, and 270°), flipping (horizontal/vertical), resizing, random amounts of translation, blurring, sharpening, adding jitters in color and/or luminance, contrasting histogram equalization, etc [47,51,52,56,60-63]. Another type of augmentation relates to the patch generation strategy; applying large medical images directly to the CNN is impractical. From a large pathological image, with a size between 1024 × 1024 (camera) and > 10⁴ × 10⁴ (scanner) pixels, smaller patches with sizes between 32 × 32 and 512 × 512 pixels are retrieved for use in training and inference of CNNs. Instead of using the pre-generated set of image patches through the whole training process, resampling patches during each training epoch can introduce more variance in training data to reduce the chance of overfitting [60].

After the patch-level CNN is trained, another ML model is often developed for the whole image level decision. In this case, a patch-level decision is made for every single patch in the training images to generate heatmap-like output, from which several features are extracted via conventional image analysis methods. Then, collected feature values for the training images are fed into the target image level ML model. An example workflow for developing and using this two-stage pathology AI is depicted in Fig. 3.

EXAMPLES OF PATHOLOGY ARTIFICIAL INTELLIGENCE

CNN-based breast cancer diagnosis was tried with fine needle aspiration (FNA) cytology images [46], optical coherence tomography (OCT) images [48], and H&E stained tissue images [49], each with varying numbers of data points and model structures. A total of 175 cytology images captured by a microscopic camera at 40 × magnification level were manually split into 918 ROIs, 256 × 256 pixels in size, where each ROI had multiple cells [46]. A CNN was trained to determine the malignancy of a given ROI, and the cytological image was classified as malignant when > 30% of the ROIs in the image were malignant. The reported accuracy was 89.7%, which was far inferior to the 99.4% accuracy of a random forest classifier with 14 hand-crafted features. In order to attempt an automated intraoperative margin assessment, 4,921 frame images from the frozen section OCT were used, from which patches 64 × 64 pixels in size were extracted, resized to 32 × 32 pixels, and used for training and evaluation [48]. Patch-level CNN performance was measured, giving an accuracy of 95.0% and AUC of 0.984 in the best setting. In another study, 2,387 H&E stained breast biopsies were scanned at a magnification of 20 × [49]. Multiple CNNs were used in this study: the first CNN classified each image pixel as fat, stroma, or epithelium; the second CNN predicted whether each stromal pixel was associated with a cancer; and the third CNN determined the whole-slide-level malignancy. The reported slide level AUC was 0.962. A notable result is that, while the CNNs were trained with stromal tissues in benign slides and invasive cancer slides only, the predicted cancer association probability of the stroma near the ductal carcinoma in situ (DCIS) lesion properly related to the severity of DCIS. CNN-based lymph node metastasis detection was also tried with a different model and dataset [47,50]. Conditional random field was adopted on top of convolutional layers in order to regulate the metastasis prediction [47]. From the whole slide images (WSIs) in the CAMELYON16 dataset [72], benign and tumor image patches 768 × 768 pixels in size were sampled to train and validate the model, giving patch-level accuracy of 93.8% after incorporating data augmentation methods. In another study, 271 WSIs scanned at a magnification of 20 × were used in developing a CNN-based model for detecting micro- or macro-metastasis-free slides [50]. Region-level annotations on training images were utilized. Slide-level metastasis detection was performed after metastasis probability map generation by patch-level CNN, incorporating probability thresholding (> 0.3) and connected component analysis to remove small lesions (< 0.02 mm diameter), resulting in a detection AUC of 0.90. Mitosis detection was tried with a CNN that decides whether the center of the given image is mitotic or not [51], trained and evaluated with 50 images from five biopsy slides containing about 300 mitoses total, adopting data augmentation techniques including patch rotation and flipping. In the evaluation, a mitosis probability map was created for the given image, and pixels with locally maximal probabilities were considered as mitotic, resulting in detection F1-score 0.782.

Automatic lung cancer subtype determination was tried with FNA cytology images and H&E stained WSIs [52,54]. A total of 298 images from 76 cases acquired using a microscopic camera at 40 × magnification level were utilized in developing a CNN receiving 256 × 256 pixel images as input; the dataset comprised 82 adenocarcinomas, 125 squamous cell carcinomas, and 91 small cell carcinomas [52]. Data augmentation techniques like rotation, flipping, and Gaussian filtering were adopted to enhance the classification accuracy from 62.1% to 71.1%. A total of 1,635 WSIs from the The Cancer Genome Atlas (TCGA) [73] dataset were used in detection of lung cancer type with CNN [54]. Each input patch (512 × 512 pixels) was classified as adenocarcinoma, squamous cell carcinoma or benign, and then the averaged probability of non-benign patches was used in the slide-level decision, resulting in slide level classification AUC of 0.97, which is much superior to the previous SVM-based approach [53]. Moreover, by using the multi-task transfer learning approach, mutations of six genes including KRAS, EGFR, and STK11 were independently able to be determined on the input WSI of lung adenocarcinoma patches. The mutation detection had an AUC of 0.86 for STK11 and an AUC of 0.83 for EGFR.

Prostate cancer diagnosis has been one of the most active fields in adopting DL because of its large dependence on tissue morphology. Prostatic tissues from various sources have been used in malignancy and severity decisions [50,55-58]. In one study, 225 prostate needle biopsy slides were scanned at 40 × magnification, and malignant regions were annotated in developing a cancer detector [50]. A CNN-based patch-level cancer detection was performed for every overlapping patch in a slide to generate a probability map, and a cumulative probability histogram was created and analyzed in slide-level malignancy determination (AUC 0.99). In another study, 12,160 needle biopsy images were utilized in developing a CNN-based slide-level malignancy detector [55]. To train a patch-classifying CNN with no patch/regionlevel manual annotation, multiple instance learning was used; with a large number of WSIs (> 8,000), the result was useful (AUC 0.98). A total of 886 tissue microarray (TMA) samples were used in a trial of automated Gleason scoring [56], where 508 TMA images for training were manually segmented into combinations of benign, Gleason pattern 3, 4, and 5; 133 TMA images were used for tuning and 245 images were used for validation. The TMA level score was determined by the two most dominant patterns measured from the per-pattern probability maps generated by a trained patch-level CNN classifier. In grading the validation set, Cohen’s kappa between two pathologists was 0.71, while those between the model and each of the two pathologists were 0.75 and 0.71. 342 cases from TCGA, teaching hospital and medical lab were utilized in training automated Gleason scoring system [58], where CNN and k-NN classifier were ensembled. A total of 912 slide images were annotated with the region level to be used in training CNN to generate a pattern map for a given slide image; 1,159 slides were used to train the k-NN classifier that determines the Gleason group for the given pattern map statistics. The reported grading accuracy measured on 331 slides was 0.70, while the average accuracy of 29 general pathologists was 0.61, which is superior to the previous TCGA-based result that showed 75% accuracy in discriminating Gleason score 3 + 4 and 4 + 3 [57].

An automated determination of brain cancer severity was tried with TCGA brain cancer data [59]. A cascade of CNNs was used: an initial CNN trained with 22 WSIs for discriminating between glioblastoma (GBM) and low-grade glioma (LGG), and a secondary CNN trained with an additional 22 WSIs for discriminating between LGG grades 2 and 3. Each H&E-stained RGB color image was transformed into an H-stained channel and an E-stained channel, and only the H-stained channel was used for further analysis. The first CNN showed GBM/LGG discrimination accuracy of 96%, but the LGG grade discrimination was not so successful (71%). Survival analysis using CNN was also tried [60]. Again, 1,061 WSIs from TCGA dataset were used. For each training epoch, 256 × 256 pixel patches were sampled from manually identified, 1,024 × 1,024 pixel ROIs. At diagnosis, ROI-wise risk was determined as the median risk of nine patches sampled from the ROI, and the sample-level risk was determined as the second highest risk among ROI risks. The measured c-index of this kind of survival analysis was 0.75, which was elevated to 0.80 by modifying the CNN to receive the mutation information at its fully connected layer.

Ovarian cancer subtype classification based on CNN was tried [61]. 7,392 images were generated by splitting and cropping the original images acquired by the microscopic camera at 40 × magnification level. Rotation and image quality enhancement were used in the data augmentation phase, which enhanced the classification accuracy from 72.8% to 78.2%. Cervical cancer diagnosis on cytological images was also tried [62]. Without cell-wise segmentation, nuclei-centered cell patches were sampled from the original cytology image, followed by augmentation operations like rotation and translation. Convolutional layer parameters that were trained by using ImageNet data were transferred to actual CNN. Herlev and HEMLBC datasets were used in evaluation, giving 98.3% and 98.6% accuracy, respectively, in five-fold cross-validation. Red blood cell (RBC) classification is crucial in sickle cell disease diagnosis. A CNN-based automatic RBC classification was tried [63], where 7,206 cell patches were generated from 434 microscopic images and used for training and testing of the classifier. Rotation and flipping were used to augment training data. Five-fold cross-validation showed an average accuracy of 89.3% in five-class coarse classification, and 87.5% in eight-class refined classification. A total of 469 TMA cores from the gastric cancer patients were used in a CNN-based survival analysis [64]. CD8 and Ki67 immunostained images were acquired and fed into separate patch-wise risk-predicting CNNs for each stain. From the differential analysis between the low-risk group and the high-risk group, it was claimed that the density of CD8 cells was largely related to the risk level.

FUTURE PROSPECTS

We have provided an overview of various medical applications of AI technology, especially in pathology. It is encouraging that the accuracy of automated morphological analyses has improved due to DL technology. The pathologic field in AI is expanding to disease severity assessment and prognosis prediction. Although most AI research in pathology is still focused on cancer detection and the grading of tumors, pathological diagnosis is not simply a morphological diagnosis, but is a complex process of evaluation and judgment of various types of clinical data that deal with various organs and diseases. A large amount of data, including genetic data, clinical data, and digital images, is needed to develop AI that covers the range of clinical situations. There are a number of public medical databases, including TCGA, and a number of studies have been done based on those databases. They provide a good starting point in researching and developing a medical AI, but it requires much more high-quality data; e.g., detailed annotations on a large number of pathology images, created and validated by several experienced pathologists, are necessary to develop a pathology-image-analyzing AI that is comparable to human pathologists.

There are difficulties in constructing such high-quality data in reality, largely due to the protection of privacy, proprietary techniques, and the lack of funding and pathologists to participate in the annotation process. To overcome this data insufficiency, as we have mentioned earlier, several techniques have been introduced, such as transfer learning and data augmentation. Still, these techniques are sub-optimal; transfer learning cannot guarantee the optimal convolutional filters specific for the task, and data augmentation cannot deal with the unseen data and patterns. The ultimate solution is to construct a large amount of thoroughly labeled and annotated medical data, through the cooperation of multiple hospitals and medical laboratories. To accelerate the construction of such a dataset, efficient tools for labeling and annotating are required, which can be assisted by another type of AI [45].

Eventually, there will be a medical AI of the prognostic prediction model, combining clinical data, genetic data, and morphology. Also, a new grading system applicable to several tumors can be created by an AI model that has learned from the patient’s prognosis combined with a number of variables including morphology, treatment modality, and tumor markers, etc. This will also help to overcome the poor reproducibility and the variety of current grading and staging results among pathologists, leading to much better clinical outcomes for patients.

Notes

Conflicts of Interest

The authors declare that they have no potential conflicts of interest.

Acknowledgments

Fig. 1.

A simplified modern convolutional neural network (CNN) architecture example. In contrast to the classic CNN comprising only a cascade of convolution layers and pooling layers followed by a few fully connected layers, this example has various other concepts like branching from the max pooling layer to several (1×1, 3×3, 5 × 5) convolution layers as well as the average pooling layer, merging by concatenation from two (1×1, 5×5) convolution layers and the average pooling layer, and residual addition of max pooling layer output to the output of its succeeding (3×3) convolution layer.

Fig. 2.

A typical recurrent layer example. In receiving a new input x_t at time t, hidden state h_t is updated based on x_t and the previous state h_t-1 first, then output y_t is generated based on h_t. At training time, parameters like U, V, W, b_h, and b_y are trained to accurately generate y_t for every time t.

Fig. 3.

An example workflow for two-stage pathology artificial intelligence. Training phase: from the collected pathology images, a proper amount of annotation data is constructed (a). Image patch sets of balanced size are used in the training of patch-level convolutional neural network (CNN). After the patch-level CNN is trained sufficiently, heatmaps are generated for another set of pathology images using that CNN, from where the features are extracted for the decision forest like image-level machine learning (ML) model training (b). Inference phase: patch-level CNN runs on every single patch in the input pathology to generate a heatmap (first stage). Features are then extracted as in the training phase, and fed into the image-level ML model to determine the image-level result (second stage).

References

• 1. McCorduck P. Machines who think: a personal inquiry into the history and prospects of artificial intelligence. Natick: A.K. Peters, 2004.
• 2. Turing AM. I. Computing machinery and intelligence. Mind 1950; 59: 433–60. ArticlePDF
• 3. Searle JR. Minds, brains, and programs. Behav Brain Sci 1980; 3: 417–24. Article
• 4. Russell SJ, Norvig P. Artificial intelligence: a modern approach. Upper Saddle River: Prentice Hall, 2003.
• 5. Artificial intelligence [Internet] Wikipedia, 2018 [cited 2018 Dec 9]. Available from: https://en.wikipedia.org/wiki/Artificial_intelligence.
• 6. Mortensen TL, Watt DL, Leistritz FL. Loan default prediction using logistic regression and a loan pricing model. Report No. 119549 [Internet] Fargo: North Dakota State University, 1988 [cited 2018 Dec 7]. Available from: https://ideas.repec.org/p/ags/nddmrs/119549.html.
• 7. Graham P. Better Bayesian filtering [Internet] PAUL GRAHAM, 2003 [cited 2018 Nov 22]. Available from: http://www.paulgraham.com/better.html.
• 8. Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. In: Pereira F, Burges CJ, Bottou L, Weinberger KQ, eds. Advances in neural information processing systems 25. Red Hook: Curran Associates, Inc, 2012; 1097–105.
• 9. Mnih V, Kavukcuoglu K, Silver D, et al. Human-level control through deep reinforcement learning. Nature 2015; 518: 529–33. ArticlePubMedPDF
• 10. Silver D, Schrittwieser J, Simonyan K, et al. Mastering the game of Go without human knowledge. Nature 2017; 550: 354–9. ArticlePubMedPDF
• 11. Hannun A, Case C, Casper J. Deep speech: scaling up end-to-end speech recognition [Internet] Ithaca: arXiv, Cornell University, 2014 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1412.5567.
• 12. Luong MT, Pham H, Manning CD. Effective approaches to attention-based neural machine translation. In: Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing; 2015 Sep 17-21; Lisbon, Portugal. Stroudsburg. Association for Computational Linguistics. 2015; 1412–21. Article
• 13. Wu Y, Schuster M, Chen Z. Google’s neural machine translation system: bridging the gap between human and machine translation [Internet] Ithaca: arXiv, Cornell University, 2016 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1609.08144.
• 14. Antol S, Agrawal A, Lu J, et al. VQA: visual question answering. In: Proceedings of the IEEE International Conference on Computer Vision; 2015 Dec 7-13; Santiago, Chile. Washington, DC. IEEE Computer Society. 2015; 2425–33. Article
• 15. Kim JH, Lee SW, Kwak D, et al. Multimodal residual learning for visual QA. In : Lee DD, von Luxburg U, Garnett R, eds. Advances in neural information processing systems 29. Red Hook: NY Curran Associates Inc, 2016; 361–9.
• 16. LeCun Y, Bottou L, Bengio Y, Haffner P. Gradient-based learning applied to document recognition. Proc IEEE 1998; 86: 2278–324. Article
• 17. Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015 Jun 7-12, Boston, MA, USA. Silver Spring: IEEE Computer Society Press, 2015; 1–9.
• 18. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature 2015; 521: 436–44. ArticlePubMedPDF
• 19. Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput 1997; 9: 1735–80. ArticlePubMed
• 20. Weizenbaum J. ELIZA: a computer program for the study of natural language communication between man and machine. Commun ACM 1966; 9: 36–45. Article
• 21. Shortliffe EH. Mycin: a knowledge-based computer program applied to infectious diseases. In: Proceedings of the Annual Symposium on Computer Application in Medical Care, 1977 Oct 3-5, Washington, DC, USA. New York: Institute of Electrical and Electronics Engineers, 1977; 66–9.
• 22. Heckerman DE, Horvitz EJ, Nathwani BN. Toward normative expert systems: Part I. The Pathfinder project. Methods Inf Med 1992; 31: 90–105. ArticlePubMedPDF
• 23. Heckerman DE, Nathwani BN. Toward normative expert systems: Part II. Probability-based representations for efficient knowledge acquisition and inference. Methods Inf Med 1992; 31: 106–16. ArticlePubMedPDF
• 24. Vyborny CJ, Giger ML. Computer vision and artificial intelligence in mammography. AJR Am J Roentgenol 1994; 162: 699–708. ArticlePubMed
• 25. Kononenko I. Machine learning for medical diagnosis: history, state of the art and perspective. Artif Intell Med 2001; 23: 89–109. ArticlePubMed
• 26. Baker JA, Rosen EL, Lo JY, Gimenez EI, Walsh R, Soo MS. Computeraided detection (CAD) in screening mammography: sensitivity of commercial CAD systems for detecting architectural distortion. AJR Am J Roentgenol 2003; 181: 1083–8. ArticlePubMed
• 27. Gulshan V, Peng L, Coram M, et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA 2016; 316: 2402–10. ArticlePubMed
• 28. Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis. Med Image Anal 2017; 42: 60–88. ArticlePubMed
• 29. Kohli M, Prevedello LM, Filice RW, Geis JR. Implementing machine learning in radiology practice and research. AJR Am J Roentgenol 2017; 208: 754–60. ArticlePubMed
• 30. Shaikhina T, Khovanova NA. Handling limited datasets with neural networks in medical applications: a small-data approach. Artif Intell Med 2017; 75: 51–63. ArticlePubMed
• 31. Angermueller C, Parnamaa T, Parts L, Stegle O. Deep learning for computational biology. Mol Syst Biol 2016; 12: 878.ArticlePubMedPMC
• 32. Torkamani A, Andersen KG, Steinhubl SR, Topol EJ. High-definition medicine. Cell 2017; 170: 828–43. ArticlePubMedPMC
• 33. Wainberg M, Merico D, Delong A, Frey BJ. Deep learning in biomedicine. Nat Biotechnol 2018; 36: 829–38. ArticlePubMedPDF
• 34. Xiong HY, Alipanahi B, Lee LJ, et al. RNA splicing: the human splicing code reveals new insights into the genetic determinants of disease. Science 2015; 347: 1254806.ArticlePubMed
• 35. Poplin R, Chang PC, Alexander D, et al. A universal SNP and smallindel variant caller using deep neural networks. Nat Biotechnol 2018; 36: 983–7. ArticlePubMedPDF
• 36. Rajkomar A, Oren E, Chen K, et al. Scalable and accurate deep learning with electronic health records. NPJ Digit Med 2018; 1: 18.ArticlePubMedPMCPDF
• 37. Fernandes K, Chicco D, Cardoso JS, Fernandes J. Supervised deep learning embeddings for the prediction of cervical cancer diagnosis. PeerJ Comput Sci 2018; 4: e154.ArticlePubMedPMCPDF
• 38. Ye JJ. Artificial intelligence for pathologists is not near: it is here: description of a prototype that can transform how we practice pathology tomorrow. Arch Pathol Lab Med 2015; 139: 929–35. ArticlePubMed
• 39. Beck JR, Salem DN, Estes NA, Pauker SG. A computer-based Markov decision analysis of the management of symptomatic bifascicular block: the threshold probability for pacing. J Am Coll Cardiol 1987; 9: 920–35. ArticlePubMed
• 40. Schaefer AJ, Bailey MD, Shechter SM, Roberts MS. Modeling medical treatment using Markov decision processes. In: Brandeau ML, Sainfort F, Pierskalla WP, eds. Operations research and health care: a handbook of methods and applications. Boston: Kluwer Academic Publisher, 2004; 593–612.
• 41. Alagoz O, Hsu H, Schaefer AJ, Roberts MS. Markov decision processes: a tool for sequential decision making under uncertainty. Med Decis Making 2010; 30: 474–83. ArticlePubMed
• 42. Harbias A, Salmo E, Crump A. Implications of observer variation in Gleason scoring of prostate cancer on clinical management: a collaborative audit. Gulf J Oncolog 2017; 1: 41–5. PubMed
• 43. Ozkan TA, Eruyar AT, Cebeci OO, Memik O, Ozcan L, Kuskonmaz I. Interobserver variability in Gleason histological grading of prostate cancer. Scand J Urol 2016; 50: 420–4. ArticlePubMed
• 44. Janowczyk A, Madabhushi A. Deep learning for digital pathology image analysis: a comprehensive tutorial with selected use cases. J Pathol Inform 2016; 7: 29.ArticlePubMedPMC
• 45. Komura D, Ishikawa S. Machine learning methods for histopathological image analysis. Comput Struct Biotechnol J 2018; 16: 34–42. ArticlePubMedPMC
• 46. Garud H, Karri SP, Sheet D, et al. High-magnification multi-views based classification of breast fine needle aspiration cytology cell samples using fusion of decisions from deep convolutional networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017 Jul 21-26, Honolulu, HI, USA. New York: Institute of Electrical and Electronics Engineers, 2017; 828–33.
• 47. Li Y, Ping W. Cancer metastasis detection with neural conditional random field [Internet] Ithaca: arXiv, Cornell University, 2018 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1806.07064.
• 48. Rannen Triki A, Blaschko MB, Jung YM, et al. Intraoperative margin assessment of human breast tissue in optical coherence tomography images using deep neural networks. Comput Med Imaging Graph 2018; 69: 21–32. ArticlePubMed
• 49. Ehteshami Bejnordi B, Mullooly M, Pfeiffer RM, et al. Using deep convolutional neural networks to identify and classify tumor-associated stroma in diagnostic breast biopsies. Mod Pathol 2018; 31: 1502–12. ArticlePubMedPMCPDF
• 50. Litjens G, Sánchez CI, Timofeeva N, et al. Deep learning as a tool for increased accuracy and efficiency of histopathological diagnosis. Sci Rep 2016; 6: 26286.ArticlePubMedPMCPDF
• 51. Cires¸an DC, Giusti A, Gambardella LM, Schmidhuber J. Mitosis detection in breast cancer histology images with deep neural networks. Med Image Comput Comput Assist Interv 2013; 16: 411–8. PubMed
• 52. Teramoto A, Tsukamoto T, Kiriyama Y, Fujita H. Automated classification of lung cancer types from cytological images using deep convolutional neural networks. Biomed Res Int 2017; 2017: 4067832.ArticlePubMedPMCPDF
• 53. Yu KH, Zhang C, Berry GJ, et al. Predicting non-small cell lung cancer prognosis by fully automated microscopic pathology image features. Nat Commun 2016; 7: 12474.ArticlePubMedPMCPDF
• 54. Coudray N, Ocampo PS, Sakellaropoulos T, et al. Classification and mutation prediction from non-small cell lung cancer histopathology images using deep learning. Nat Med 2018; 24: 1559–67. ArticlePubMedPMCPDF
• 55. Campanella G, Silva VW, Fuchs TJ. Terabyte-scale deep multiple instance learning for classification and localization in pathology [Internet] Ithaca: arXiv, Cornell University, 2018 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1805.06983.
• 56. Arvaniti E, Fricker KS, Moret M, et al. Automated Gleason grading of prostate cancer tissue microarrays via deep learning. Sci Rep 2018; 8: 12054.ArticlePubMedPMCPDF
• 57. Zhou N, Fedorov A, Fennessy F, Kikinis R, Gao Y. Large scale digital prostate pathology image analysis combining feature extraction and deep neural network [Internet] Ithaca: arXiv, Cornell University, 2017 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1705.02678.
• 58. Nagpal K, Foote D, Liu Y, et al. Development and validation of a deep learning algorithm for improving Gleason scoring of prostate cancer [Internet] Ithaca: arXiv, Cornell University, 2018 [cited 2018 Nov 22]. Available from: http://arxiv.org/abs/1811.06497.
• 59. Ertosun MG, Rubin DL. Automated grading of gliomas using deep learning in digital pathology images: a modular approach with ensemble of convolutional neural networks. AMIA Annu Symp Proc 2015; 2015: 1899–908. PubMedPMC
• 60. Mobadersany P, Yousefi S, Amgad M, et al. Predicting cancer outcomes from histology and genomics using convolutional networks. Proc Natl Acad Sci U S A 2018; 115: E2970–E9. ArticlePubMedPMC
• 61. Wu M, Yan C, Liu H, Liu Q. Automatic classification of ovarian cancer types from cytological images using deep convolutional neural networks. Biosci Rep 2018; 38: BSR20180289.ArticlePubMedPMCPDF
• 62. Zhang L, Lu L, Nogues I, Summers RM, Liu S, Yao J. DeepPap: deep convolutional networks for cervical cell classification. IEEE J Biomed Health Inform 2017; 21: 1633–43. ArticlePubMed
• 63. Xu M, Papageorgiou DP, Abidi SZ, Dao M, Zhao H, Karniadakis GE. A deep convolutional neural network for classification of red blood cells in sickle cell anemia. PLoS Comput Biol 2017; 13: e1005746. ArticlePubMedPMC
• 64. Meier A, Nekolla K, Earle S, et al. End-to-end learning to predict survival in patients with gastric cancer using convolutional neural networks. Ann Oncol 2018; 29(Suppl 8):mdy269.075. ArticlePDF
• 65. Xie W, Noble JA, Zisserman A. Microscopy cell counting and detection with fully convolutional regression networks. Comput Methods Biomech Biomed Eng Imaging Vis 2016; 6: 283–92. Article
• 66. Tuominen VJ, Ruotoistenmaki S, Viitanen A, Jumppanen M, Isola J. ImmunoRatio: a publicly available web application for quantitative image analysis of estrogen receptor (ER), progesterone receptor (PR), and Ki-67. Breast Cancer Res 2010; 12: R56.ArticlePubMedPMCPDF
• 67. Meijering E. Cell segmentation: 50 years down the road [life sciences]. IEEE Signal Process Mag 2012; 29: 140–5. Article
• 68. Ruifrok AC, Johnston DA. Quantification of histochemical staining by color deconvolution. Anal Quant Cytol Histol 2001; 23: 291–9. PubMed
• 69. Otsu N. A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern 1979; 9: 62–6. Article
• 70. Zhang L, Sonka M, Lu L, Summers RM, Yao J. Combining fully convolutional networks and graph-based approach for automated segmentation of cervical cell nuclei. In: 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), 2017 Apr 18-21, Melbourne, VIC, Australia. New York: Institute of Electrical and Electronics Engineers, 2017; 406–9.
• 71. Chen H, Qi X, Yu L, Heng PA. DCAN: deep contour-aware networks for accurate gland segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016 Jun 27-30, Las Vegas, NV, USA. New York: Institute of Electrical and Electronics Engineers, 2016; 2487–96.
• 72. CAMELYON16 Consortium. CAMELYON16. CAMELYON16 ISBI challenge on cancer metastasis detection in lymph node, 2015 [Internet] Grand-Challenges, 2016 [cited 2018 Nov 22]. Available from: https://camelyon16.grand-challenge.org/.
• 73. The Cancer Genome Atlas [Internet] Bethesda: The Cancer Genome Atlas, National Cancer Institute, 2011 [cited 2018 Nov 22]. Available from: https://cancergenome.nih.gov/.

Figure & Data

References

Citations

Citations to this article as recorded by  

• ChatGPT as an aid for pathological diagnosis of cancer
  Shaivy Malik, Sufian Zaheer
  Pathology - Research and Practice.2024; 253: 154989.     CrossRef
• Computational pathology: A survey review and the way forward
  Mahdi S. Hosseini, Babak Ehteshami Bejnordi, Vincent Quoc-Huy Trinh, Lyndon Chan, Danial Hasan, Xingwen Li, Stephen Yang, Taehyo Kim, Haochen Zhang, Theodore Wu, Kajanan Chinniah, Sina Maghsoudlou, Ryan Zhang, Jiadai Zhu, Samir Khaki, Andrei Buin, Fatemeh
  Journal of Pathology Informatics.2024; 15: 100357.     CrossRef
• Applications of artificial intelligence in the field of oral and maxillofacial pathology: a systematic review and meta-analysis
  Nishath Sayed Abdul, Ganiga Channaiah Shivakumar, Sunila Bukanakere Sangappa, Marco Di Blasio, Salvatore Crimi, Marco Cicciù, Giuseppe Minervini
  BMC Oral Health.2024;[Epub]     CrossRef
• The Constrained-Disorder Principle Assists in Overcoming Significant Challenges in Digital Health: Moving from “Nice to Have” to Mandatory Systems
  Noa Hurvitz, Yaron Ilan
  Clinics and Practice.2023; 13(4): 994.     CrossRef
• Building a nonclinical pathology laboratory of the future for pharmaceutical research excellence
  D.G. Rudmann, L. Bertrand, A. Zuraw, J. Deiters, M. Staup, Y. Rivenson, J. Kuklyte
  Drug Discovery Today.2023; 28(10): 103747.     CrossRef
• Automated image analysis of keratin 7 staining can predict disease outcome in primary sclerosing cholangitis
  Nelli Sjöblom, Sonja Boyd, Anniina Manninen, Sami Blom, Anna Knuuttila, Martti Färkkilä, Johanna Arola
  Hepatology Research.2023; 53(4): 322.     CrossRef
• Application of convolutional neural network for analyzing hepatic fibrosis in mice
  Hyun-Ji Kim, Eun Bok Baek, Ji-Hee Hwang, Minyoung Lim, Won Hoon Jung, Myung Ae Bae, Hwa-Young Son, Jae-Woo Cho
  Journal of Toxicologic Pathology.2023; 36(1): 21.     CrossRef
• Machine Learning Techniques for Prognosis Estimation and Knowledge Discovery From Lab Test Results With Application to the COVID-19 Emergency
  Alfonso Emilio Gerevini, Roberto Maroldi, Matteo Olivato, Luca Putelli, Ivan Serina
  IEEE Access.2023; 11: 83905.     CrossRef
• Artificial intelligence in dentistry—A review
  Hao Ding, Jiamin Wu, Wuyuan Zhao, Jukka P. Matinlinna, Michael F. Burrow, James K. H. Tsoi
  Frontiers in Dental Medicine.2023;[Epub]     CrossRef
• Dental Age Estimation Using the Demirjian Method: Statistical Analysis Using Neural Networks
  Byung-Yoon Roh, Jong-Seok Lee, Sang-Beom Lim, Hye-Won Ryu, Su-Jeong Jeon, Ju-Heon Lee, Yo-Seob Seo, Ji-Won Ryu, Jong-Mo Ahn
  Korean Journal of Legal Medicine.2023; 47(1): 1.     CrossRef
• The use of artificial intelligence in health care. Problems of identification of patients' conditions in the processes of detailing the diagnosis
  Mintser O
  Artificial Intelligence.2023; 28(AI.2023.28): 8.     CrossRef
• The Effectiveness of Data Augmentation for Mature White Blood Cell Image Classification in Deep Learning — Selection of an Optimal Technique for Hematological Morphology Recognition —
  Hiroyuki NOZAKA, Kosuke KAMATA, Kazufumi YAMAGATA
  IEICE Transactions on Information and Systems.2023; E106.D(5): 707.     CrossRef
• Rectal Cancer Stages T2 and T3 Identification Based on Asymptotic Hybrid Feature Maps
  Shujing Sun, Jiale Wu, Jian Yao, Yang Cheng, Xin Zhang, Zhihua Lu, Pengjiang Qian
  Computer Modeling in Engineering & Sciences.2023; 137(1): 923.     CrossRef
• How to use AI in pathology
  Peter Schüffler, Katja Steiger, Wilko Weichert
  Genes, Chromosomes and Cancer.2023; 62(9): 564.     CrossRef
• Cutting-Edge Technologies for Digital Therapeutics: A Review and Architecture Proposals for Future Directions
  Joo Hun Yoo, Harim Jeong, Tai-Myoung Chung
  Applied Sciences.2023; 13(12): 6929.     CrossRef
• A convolutional neural network STIFMap reveals associations between stromal stiffness and EMT in breast cancer
  Connor Stashko, Mary-Kate Hayward, Jason J. Northey, Neil Pearson, Alastair J. Ironside, Johnathon N. Lakins, Roger Oria, Marie-Anne Goyette, Lakyn Mayo, Hege G. Russnes, E. Shelley Hwang, Matthew L. Kutys, Kornelia Polyak, Valerie M. Weaver
  Nature Communications.2023;[Epub]     CrossRef
• Artificial Intelligence-Based PTEN Loss Assessment as an Early Predictor of Prostate Cancer Metastasis After Surgery: A Multicenter Retrospective Study
  Palak Patel, Stephanie Harmon, Rachael Iseman, Olga Ludkowski, Heidi Auman, Sarah Hawley, Lisa F. Newcomb, Daniel W. Lin, Peter S. Nelson, Ziding Feng, Hilary D. Boyer, Maria S. Tretiakova, Larry D. True, Funda Vakar-Lopez, Peter R. Carroll, Matthew R. Co
  Modern Pathology.2023; 36(10): 100241.     CrossRef
• Minimum resolution requirements of digital pathology images for accurate classification
  Lydia Neary-Zajiczek, Linas Beresna, Benjamin Razavi, Vijay Pawar, Michael Shaw, Danail Stoyanov
  Medical Image Analysis.2023; 89: 102891.     CrossRef
• Whole Slide Imaging Technology and Its Applications: Current and Emerging Perspectives
  Ekta Jain, Ankush Patel, Anil V. Parwani, Saba Shafi, Zoya Brar, Shivani Sharma, Sambit K. Mohanty
  International Journal of Surgical Pathology.2023;[Epub]     CrossRef
• Artificial Intelligence in the Pathology of Gastric Cancer
  Sangjoon Choi, Seokhwi Kim
  Journal of Gastric Cancer.2023; 23(3): 410.     CrossRef
• Endoscopic Ultrasound-Based Artificial Intelligence Diagnosis of Pancreatic Cystic Neoplasms
  Jin-Seok Park, Seok Jeong
  The Korean Journal of Pancreas and Biliary Tract.2023; 28(3): 53.     CrossRef
• Framework for Classifying Explainable Artificial Intelligence (XAI) Algorithms in Clinical Medicine
  Thomas Gniadek, Jason Kang, Talent Theparee, Jacob Krive
  Online Journal of Public Health Informatics.2023; 15: e50934.     CrossRef
• A Literature Review of the Future of Oral Medicine and Radiology, Oral Pathology, and Oral Surgery in the Hands of Technology
  Ishita Singhal, Geetpriya Kaur, Dirk Neefs, Aparna Pathak
  Cureus.2023;[Epub]     CrossRef
• AI-Powered Biomolecular-Specific and Label-Free Multispectral Imaging Rapidly Detects Malignant Neoplasm in Surgically Excised Breast Tissue Specimens
  Rishikesh Pandey, David Fournier, Gary Root, Machele Riccio, Aditya Shirvalkar, Gianfranco Zamora, Noel Daigneault, Michael Sapack, Minghao Zhong, Malini Harigopal
  Archives of Pathology & Laboratory Medicine.2023; 147(11): 1298.     CrossRef
• Exploring the status of artificial intelligence for healthcare research in Africa: a bibliometric and thematic analysis
  Tabu S. Kondo, Salim A. Diwani, Ally S. Nyamawe, Mohamed M. Mjahidi
  AI and Ethics.2023;[Epub]     CrossRef
• Artificial intelligence for patient scheduling in the real-world health care setting: A metanarrative review
  Dacre R.T. Knight, Christopher A. Aakre, Christopher V. Anstine, Bala Munipalli, Parisa Biazar, Ghada Mitri, Jose Raul Valery, Tara Brigham, Shehzad K. Niazi, Adam I. Perlman, John D. Halamka, Abd Moain Abu Dabrh
  Health Policy and Technology.2023; 12(4): 100824.     CrossRef
• Towards Autonomous Healthcare: Integrating Artificial Intelligence (AI) for Personalized Medicine and Disease Prediction
  Nitin Rane, Saurabh Choudhary, Jayesh Rane
  SSRN Electronic Journal.2023;[Epub]     CrossRef
• Medical imaging and multimodal artificial intelligence models for streamlining and enhancing cancer care: opportunities and challenges
  Kevin Pierre, Manas Gupta, Abheek Raviprasad, Seyedeh Mehrsa Sadat Razavi, Anjali Patel, Keith Peters, Bruno Hochhegger, Anthony Mancuso, Reza Forghani
  Expert Review of Anticancer Therapy.2023; 23(12): 1265.     CrossRef
• Automated differential diagnostics of respiratory diseases using an electronic stethoscope
  Diana Arhypenko, Denis Panaskin, Dmytro Babko
  Polish Journal of Medical Physics and Engineering.2023; 29(4): 208.     CrossRef
• Automated abstraction of myocardial perfusion imaging reports using natural language processing
  Parija Sharedalal, Ajay Singh, Neal Shah, Diwakar Jain
  Journal of Nuclear Cardiology.2022; 29(3): 1188.     CrossRef
• Polyploid giant cancer cell characterization: New frontiers in predicting response to chemotherapy in breast cancer
  Geetanjali Saini, Shriya Joshi, Chakravarthy Garlapati, Hongxiao Li, Jun Kong, Jayashree Krishnamurthy, Michelle D. Reid, Ritu Aneja
  Seminars in Cancer Biology.2022; 81: 220.     CrossRef
• A Comprehensive Review of Markov Random Field and Conditional Random Field Approaches in Pathology Image Analysis
  Yixin Li, Chen Li, Xiaoyan Li, Kai Wang, Md Mamunur Rahaman, Changhao Sun, Hao Chen, Xinran Wu, Hong Zhang, Qian Wang
  Archives of Computational Methods in Engineering.2022; 29(1): 609.     CrossRef
• Artificial intelligence in oncology: From bench to clinic
  Jamal Elkhader, Olivier Elemento
  Seminars in Cancer Biology.2022; 84: 113.     CrossRef
• Yeast‐like organisms phagocytosed by circulating neutrophils: Evidence of disseminated histoplasmosis
  Yue Zhao, Jenna McCracken, Endi Wang
  International Journal of Laboratory Hematology.2022; 44(1): 51.     CrossRef
• Whole-slide imaging, tissue image analysis, and artificial intelligence in veterinary pathology: An updated introduction and review
  Aleksandra Zuraw, Famke Aeffner
  Veterinary Pathology.2022; 59(1): 6.     CrossRef
• A comprehensive review of computer-aided whole-slide image analysis: from datasets to feature extraction, segmentation, classification and detection approaches
  Xintong Li, Chen Li, Md Mamunur Rahaman, Hongzan Sun, Xiaoqi Li, Jian Wu, Yudong Yao, Marcin Grzegorzek
  Artificial Intelligence Review.2022; 55(6): 4809.     CrossRef
• Liquid Biopsy and Artificial Intelligence as Tools to Detect Signatures of Colorectal Malignancies: A Modern Approach in Patient’s Stratification
  Octav Ginghina, Ariana Hudita, Marius Zamfir, Andrada Spanu, Mara Mardare, Irina Bondoc, Laura Buburuzan, Sergiu Emil Georgescu, Marieta Costache, Carolina Negrei, Cornelia Nitipir, Bianca Galateanu
  Frontiers in Oncology.2022;[Epub]     CrossRef
• Automated bone marrow cytology using deep learning to generate a histogram of cell types
  Rohollah Moosavi Tayebi, Youqing Mu, Taher Dehkharghanian, Catherine Ross, Monalisa Sur, Ronan Foley, Hamid R. Tizhoosh, Clinton J. V. Campbell
  Communications Medicine.2022;[Epub]     CrossRef
• Risultati di esami di laboratorio per intelligenza artificiale e "machine learning"
  Marco PRADELLA
  La Rivista Italiana della Medicina di Laboratorio.2022;[Epub]     CrossRef
• The Deception of Certainty: how Non-Interpretable Machine Learning Outcomes Challenge the Epistemic Authority of Physicians. A deliberative-relational Approach
  Florian Funer
  Medicine, Health Care and Philosophy.2022; 25(2): 167.     CrossRef
• Deep discriminative learning model with calibrated attention map for the automated diagnosis of diffuse large B-cell lymphoma
  Sautami Basu, Ravinder Agarwal, Vishal Srivastava
  Biomedical Signal Processing and Control.2022; 76: 103728.     CrossRef
• Question and Answer Techniques for Financial Audits in Universities Based on Deep Learning
  Qiang Li, Hangjun Che
  Mathematical Problems in Engineering.2022; 2022: 1.     CrossRef
• Noninvasive Screening Tool for Hyperkalemia Using a Single-Lead Electrocardiogram and Deep Learning: Development and Usability Study
  Erdenebayar Urtnasan, Jung Hun Lee, Byungjin Moon, Hee Young Lee, Kyuhee Lee, Hyun Youk
  JMIR Medical Informatics.2022; 10(6): e34724.     CrossRef
• Impact of artificial intelligence on pathologists’ decisions: an experiment
  Julien Meyer, April Khademi, Bernard Têtu, Wencui Han, Pria Nippak, David Remisch
  Journal of the American Medical Informatics Association.2022; 29(10): 1688.     CrossRef
• Rapid Screening Using Pathomorphologic Interpretation to Detect BRAFV600E Mutation and Microsatellite Instability in Colorectal Cancer
  Satoshi Fujii, Daisuke Kotani, Masahiro Hattori, Masato Nishihara, Toshihide Shikanai, Junji Hashimoto, Yuki Hama, Takuya Nishino, Mizuto Suzuki, Ayatoshi Yoshidumi, Makoto Ueno, Yoshito Komatsu, Toshiki Masuishi, Hiroki Hara, Taito Esaki, Yoshiaki Nakamu
  Clinical Cancer Research.2022; 28(12): 2623.     CrossRef
• Using Deep Learning to Predict Final HER2 Status in Invasive Breast Cancers That are Equivocal (2+) by Immunohistochemistry
  Sean A. Rasmussen, Valerie J. Taylor, Alexi P. Surette, Penny J. Barnes, Gillian C. Bethune
  Applied Immunohistochemistry & Molecular Morphology.2022; 30(10): 668.     CrossRef
• Deep Neural Network for the Prediction of KRAS Genotype in Rectal Cancer
  Waleed M Ghareeb, Eman Draz, Khaled Madbouly, Ahmed H Hussein, Mohammed Faisal, Wagdi Elkashef, Mona Hany Emile, Marcus Edelhamre, Seon Hahn Kim, Sameh Hany Emile
  Journal of the American College of Surgeons.2022; 235(3): 482.     CrossRef
• Next Generation Digital Pathology: Emerging Trends and Measurement Challenges for Molecular Pathology
  Alex Dexter, Dimitrios Tsikritsis, Natalie A. Belsey, Spencer A. Thomas, Jenny Venton, Josephine Bunch, Marina Romanchikova
  Journal of Molecular Pathology.2022; 3(3): 168.     CrossRef
• Animation Design of Multisensor Data Fusion Based on Optimized AVOD Algorithm
  Li Ding, Guobing Wei, Kai Zhang, Gengxin Sun
  Journal of Sensors.2022; 2022: 1.     CrossRef
• Study on Machine Translation Teaching Model Based on Translation Parallel Corpus and Exploitation for Multimedia Asian Information Processing
  Yan Gong
  ACM Transactions on Asian and Low-Resource Language Information Processing.2022;[Epub]     CrossRef
• Analysis and Estimation of Pathological Data and Findings with Deep Learning Methods
  Ahmet Anıl ŞAKIR, Ali Hakan IŞIK, Özlem ÖZMEN, Volkan İPEK
  Veterinary Journal of Mehmet Akif Ersoy University.2022; 7(3): 175.     CrossRef
• Artificial Intelligence in Pathology: Friend or Enemy?
  Selim Sevim, Ezgi Dicle Serbes, Murat Bahadır, Mustafa Said Kartal, Serpil Dizbay Sak
  Journal of Ankara University Faculty of Medicine.2022; 75(1): 13.     CrossRef
• Evaluation Challenges in the Validation of B7-H3 as Oral Tongue Cancer Prognosticator
  Meri Sieviläinen, Anna Maria Wirsing, Aini Hyytiäinen, Rabeia Almahmoudi, Priscila Rodrigues, Inger-Heidi Bjerkli, Pirjo Åström, Sanna Toppila-Salmi, Timo Paavonen, Ricardo D. Coletta, Elin Hadler-Olsen, Tuula Salo, Ahmed Al-Samadi
  Head and Neck Pathology.2021; 15(2): 469.     CrossRef
• Amsterdam International Consensus Meeting: tumor response scoring in the pathology assessment of resected pancreatic cancer after neoadjuvant therapy
  Boris V. Janssen, Faik Tutucu, Stijn van Roessel, Volkan Adsay, Olca Basturk, Fiona Campbell, Claudio Doglioni, Irene Esposito, Roger Feakins, Noriyoshi Fukushima, Anthony J. Gill, Ralph H. Hruban, Jeffrey Kaplan, Bas Groot Koerkamp, Seung-Mo Hong, Alyssa
  Modern Pathology.2021; 34(1): 4.     CrossRef
• Fabrication of ultra-thin 2D covalent organic framework nanosheets and their application in functional electronic devices
  Weikang Wang, Weiwei Zhao, Haotian Xu, Shujuan Liu, Wei Huang, Qiang Zhao
  Coordination Chemistry Reviews.2021; 429: 213616.     CrossRef
• Generalizability of Deep Learning System for the Pathologic Diagnosis of Various Cancers
  Hyun-Jong Jang, In Hye Song, Sung Hak Lee
  Applied Sciences.2021; 11(2): 808.     CrossRef
• Integrated digital pathology at scale: A solution for clinical diagnostics and cancer research at a large academic medical center
  Peter J Schüffler, Luke Geneslaw, D Vijay K Yarlagadda, Matthew G Hanna, Jennifer Samboy, Evangelos Stamelos, Chad Vanderbilt, John Philip, Marc-Henri Jean, Lorraine Corsale, Allyne Manzo, Neeraj H G Paramasivam, John S Ziegler, Jianjiong Gao, Juan C Peri
  Journal of the American Medical Informatics Association.2021; 28(9): 1874.     CrossRef
• Translational Applications of Artificial Intelligence and Machine Learning for Diagnostic Pathology in Lymphoid Neoplasms: A Comprehensive and Evolutive Analysis
  Julia Moran-Sanchez, Antonio Santisteban-Espejo, Miguel Angel Martin-Piedra, Jose Perez-Requena, Marcial Garcia-Rojo
  Biomolecules.2021; 11(6): 793.     CrossRef
• Development and operation of a digital platform for sharing pathology image data
  Yunsook Kang, Yoo Jung Kim, Seongkeun Park, Gun Ro, Choyeon Hong, Hyungjoon Jang, Sungduk Cho, Won Jae Hong, Dong Un Kang, Jonghoon Chun, Kyoungbun Lee, Gyeong Hoon Kang, Kyoung Chul Moon, Gheeyoung Choe, Kyu Sang Lee, Jeong Hwan Park, Won-Ki Jeong, Se Yo
  BMC Medical Informatics and Decision Making.2021;[Epub]     CrossRef
• Sliding window based deep ensemble system for breast cancer classification
  Amin Alqudah, Ali Mohammad Alqudah
  Journal of Medical Engineering & Technology.2021; 45(4): 313.     CrossRef
• Artificial intelligence and computational pathology
  Miao Cui, David Y. Zhang
  Laboratory Investigation.2021; 101(4): 412.     CrossRef
• Effects of Image Quantity and Image Source Variation on Machine Learning Histology Differential Diagnosis Models
  Elham Vali-Betts, Kevin J. Krause, Alanna Dubrovsky, Kristin Olson, John Paul Graff, Anupam Mitra, Ananya Datta-Mitra, Kenneth Beck, Aristotelis Tsirigos, Cynthia Loomis, Antonio Galvao Neto, Esther Adler, Hooman H. Rashidi
  Journal of Pathology Informatics.2021; 12(1): 5.     CrossRef
• Feasibility of deep learning‐based fully automated classification of microsatellite instability in tissue slides of colorectal cancer
  Sung Hak Lee, In Hye Song, Hyun‐Jong Jang
  International Journal of Cancer.2021; 149(3): 728.     CrossRef
• Artificial intelligence in healthcare
  Yamini D Shah, Shailvi M Soni, Manish P Patel
  Indian Journal of Pharmacy and Pharmacology.2021; 8(2): 102.     CrossRef
• Proof of Concept for a Deep Learning Algorithm for Identification and Quantification of Key Microscopic Features in the Murine Model of DSS-Induced Colitis
  Agathe Bédard, Thomas Westerling-Bui, Aleksandra Zuraw
  Toxicologic Pathology.2021; 49(4): 897.     CrossRef
• An empirical analysis of machine learning frameworks for digital pathology in medical science
  S.K.B. Sangeetha, R Dhaya, Dhruv T Shah, R Dharanidharan, K. Praneeth Sai Reddy
  Journal of Physics: Conference Series.2021; 1767(1): 012031.     CrossRef
• Application of Single-Cell Approaches to Study Myeloproliferative Neoplasm Biology
  Daniel Royston, Adam J. Mead, Bethan Psaila
  Hematology/Oncology Clinics of North America.2021; 35(2): 279.     CrossRef
• Idiosyncratic Drug-Induced Liver Injury (DILI) and Herb-Induced Liver Injury (HILI): Diagnostic Algorithm Based on the Quantitative Roussel Uclaf Causality Assessment Method (RUCAM)
  Rolf Teschke, Gaby Danan
  Diagnostics.2021; 11(3): 458.     CrossRef
• Searching Images for Consensus
  Hamid R. Tizhoosh, Phedias Diamandis, Clinton J.V. Campbell, Amir Safarpoor, Shivam Kalra, Danial Maleki, Abtin Riasatian, Morteza Babaie
  The American Journal of Pathology.2021; 191(10): 1702.     CrossRef
• Automated Classification and Segmentation in Colorectal Images Based on Self-Paced Transfer Network
  Yao Yao, Shuiping Gou, Ru Tian, Xiangrong Zhang, Shuixiang He, Zhiguo Zhou
  BioMed Research International.2021; 2021: 1.     CrossRef
• Artificial intelligence and sleep: Advancing sleep medicine
  Nathaniel F. Watson, Christopher R. Fernandez
  Sleep Medicine Reviews.2021; 59: 101512.     CrossRef
• Prospective Of Artificial Intelligence: Emerging Trends In Modern Biosciences Research
  Pradeep Kumar, Ajit Kumar Singh Yadav, Abhishek Singh
  IOP Conference Series: Materials Science and Engineering.2021; 1020(1): 012008.     CrossRef
• Use and Control of Artificial Intelligence in Patients Across the Medical Workflow: Single-Center Questionnaire Study of Patient Perspectives
  Simon Lennartz, Thomas Dratsch, David Zopfs, Thorsten Persigehl, David Maintz, Nils Große Hokamp, Daniel Pinto dos Santos
  Journal of Medical Internet Research.2021; 23(2): e24221.     CrossRef
• HEAL: an automated deep learning framework for cancer histopathology image analysis
  Yanan Wang, Nicolas Coudray, Yun Zhao, Fuyi Li, Changyuan Hu, Yao-Zhong Zhang, Seiya Imoto, Aristotelis Tsirigos, Geoffrey I Webb, Roger J Daly, Jiangning Song, Zhiyong Lu
  Bioinformatics.2021; 37(22): 4291.     CrossRef
• A Review of Applications of Artificial Intelligence in Gastroenterology
  Khalid Nawab, Ravi Athwani, Awais Naeem, Muhammad Hamayun, Momna Wazir
  Cureus.2021;[Epub]     CrossRef
• Evaluating Cancer-Related Biomarkers Based on Pathological Images: A Systematic Review
  Xiaoliang Xie, Xulin Wang, Yuebin Liang, Jingya Yang, Yan Wu, Li Li, Xin Sun, Pingping Bing, Binsheng He, Geng Tian, Xiaoli Shi
  Frontiers in Oncology.2021;[Epub]     CrossRef
• Deep learning-based histopathological segmentation for whole slide images of colorectal cancer in a compressed domain
  Hyeongsub Kim, Hongjoon Yoon, Nishant Thakur, Gyoyeon Hwang, Eun Jung Lee, Chulhong Kim, Yosep Chong
  Scientific Reports.2021;[Epub]     CrossRef
• Deep Learning on Oral Squamous Cell Carcinoma Ex Vivo Fluorescent Confocal Microscopy Data: A Feasibility Study
  Veronika Shavlokhova, Sameena Sandhu, Christa Flechtenmacher, Istvan Koveshazi, Florian Neumeier, Víctor Padrón-Laso, Žan Jonke, Babak Saravi, Michael Vollmer, Andreas Vollmer, Jürgen Hoffmann, Michael Engel, Oliver Ristow, Christian Freudlsperger
  Journal of Clinical Medicine.2021; 10(22): 5326.     CrossRef
• A Pathologist-Annotated Dataset for Validating Artificial Intelligence: A Project Description and Pilot Study
  Sarah N. Dudgeon, Si Wen, Matthew G. Hanna, Rajarsi Gupta, Mohamed Amgad, Manasi Sheth, Hetal Marble, Richard Huang, Markus D. Herrmann, Clifford H. Szu, Darick Tong, Bruce Werness, Evan Szu, Denis Larsimont, Anant Madabhushi, Evangelos Hytopoulos, Weijie
  Journal of Pathology Informatics.2021; 12(1): 45.     CrossRef
• Artificial Intelligence in Medicine: A Multinational Multi-Center Survey on the Medical and Dental Students' Perception
  Sotirios Bisdas, Constantin-Cristian Topriceanu, Zosia Zakrzewska, Alexandra-Valentina Irimia, Loizos Shakallis, Jithu Subhash, Maria-Madalina Casapu, Jose Leon-Rojas, Daniel Pinto dos Santos, Dilys Miriam Andrews, Claudia Zeicu, Ahmad Mohammad Bouhuwaish
  Frontiers in Public Health.2021;[Epub]     CrossRef
• Digital/Computational Technology for Molecular Cytology Testing: A Short Technical Note with Literature Review
  Robert Y. Osamura, Naruaki Matsui, Masato Kawashima, Hiroyasu Saiga, Maki Ogura, Tomoharu Kiyuna
  Acta Cytologica.2021; 65(4): 342.     CrossRef
• Advances in Digital Pathology: From Artificial Intelligence to Label-Free Imaging
  Frederik Großerueschkamp, Hendrik Jütte, Klaus Gerwert, Andrea Tannapfel
  Visceral Medicine.2021; 37(6): 482.     CrossRef
• Feasibility of fully automated classification of whole slide images based on deep learning
  Kyung-Ok Cho, Sung Hak Lee, Hyun-Jong Jang
  The Korean Journal of Physiology & Pharmacology.2020; 24(1): 89.     CrossRef
• Same same but different: A Web‐based deep learning application revealed classifying features for the histopathologic distinction of cortical malformations
  Joshua Kubach, Angelika Muhlebner‐Fahrngruber, Figen Soylemezoglu, Hajime Miyata, Pitt Niehusmann, Mrinalini Honavar, Fabio Rogerio, Se‐Hoon Kim, Eleonora Aronica, Rita Garbelli, Samuel Vilz, Alexander Popp, Stefan Walcher, Christoph Neuner, Michael Schol
  Epilepsia.2020; 61(3): 421.     CrossRef
• Segmentation and Classification in Digital Pathology for Glioma Research: Challenges and Deep Learning Approaches
  Tahsin Kurc, Spyridon Bakas, Xuhua Ren, Aditya Bagari, Alexandre Momeni, Yue Huang, Lichi Zhang, Ashish Kumar, Marc Thibault, Qi Qi, Qian Wang, Avinash Kori, Olivier Gevaert, Yunlong Zhang, Dinggang Shen, Mahendra Khened, Xinghao Ding, Ganapathy Krishnamu
  Frontiers in Neuroscience.2020;[Epub]     CrossRef
• Artificial intelligence as the next step towards precision pathology
  B. Acs, M. Rantalainen, J. Hartman
  Journal of Internal Medicine.2020; 288(1): 62.     CrossRef
• Introduction to digital pathology and computer-aided pathology
  Soojeong Nam, Yosep Chong, Chan Kwon Jung, Tae-Yeong Kwak, Ji Youl Lee, Jihwan Park, Mi Jung Rho, Heounjeong Go
  Journal of Pathology and Translational Medicine.2020; 54(2): 125.     CrossRef
• Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine
  Zeeshan Ahmed, Khalid Mohamed, Saman Zeeshan, XinQi Dong
  Database.2020;[Epub]     CrossRef
• Scoring pleurisy in slaughtered pigs using convolutional neural networks
  Abigail R. Trachtman, Luca Bergamini, Andrea Palazzi, Angelo Porrello, Andrea Capobianco Dondona, Ercole Del Negro, Andrea Paolini, Giorgio Vignola, Simone Calderara, Giuseppe Marruchella
  Veterinary Research.2020;[Epub]     CrossRef
• Current Status of Computational Intelligence Applications in Dermatological Clinical Practice
  Carmen Rodríguez-Cerdeira, José Luís González-Cespón, Roberto Arenas
  The Open Dermatology Journal.2020; 14(1): 6.     CrossRef
• New unified insights on deep learning in radiological and pathological images: Beyond quantitative performances to qualitative interpretation
  Yoichi Hayashi
  Informatics in Medicine Unlocked.2020; 19: 100329.     CrossRef
• Artificial Intelligence in Cardiology: Present and Future
  Francisco Lopez-Jimenez, Zachi Attia, Adelaide M. Arruda-Olson, Rickey Carter, Panithaya Chareonthaitawee, Hayan Jouni, Suraj Kapa, Amir Lerman, Christina Luong, Jose R. Medina-Inojosa, Peter A. Noseworthy, Patricia A. Pellikka, Margaret M. Redfield, Vero
  Mayo Clinic Proceedings.2020; 95(5): 1015.     CrossRef
• Artificial intelligence in oncology
  Hideyuki Shimizu, Keiichi I. Nakayama
  Cancer Science.2020; 111(5): 1452.     CrossRef
• Artificial intelligence and the future of global health
  Nina Schwalbe, Brian Wahl
  The Lancet.2020; 395(10236): 1579.     CrossRef
• The future of pathology is digital
  J.D. Pallua, A. Brunner, B. Zelger, M. Schirmer, J. Haybaeck
  Pathology - Research and Practice.2020; 216(9): 153040.     CrossRef
• Weakly-supervised learning for lung carcinoma classification using deep learning
  Fahdi Kanavati, Gouji Toyokawa, Seiya Momosaki, Michael Rambeau, Yuka Kozuma, Fumihiro Shoji, Koji Yamazaki, Sadanori Takeo, Osamu Iizuka, Masayuki Tsuneki
  Scientific Reports.2020;[Epub]     CrossRef
• The use of artificial intelligence, machine learning and deep learning in oncologic histopathology
  Ahmed S. Sultan, Mohamed A. Elgharib, Tiffany Tavares, Maryam Jessri, John R. Basile
  Journal of Oral Pathology & Medicine.2020; 49(9): 849.     CrossRef
• Convergence of Digital Pathology and Artificial Intelligence Tools in Anatomic Pathology Practice: Current Landscape and Future Directions
  Anil V. Parwani, Mahul B. Amin
  Advances in Anatomic Pathology.2020; 27(4): 221.     CrossRef
• Advances in tissue-based imaging: impact on oncology research and clinical practice
  Arman Rahman, Chowdhury Jahangir, Seodhna M. Lynch, Nebras Alattar, Claudia Aura, Niamh Russell, Fiona Lanigan, William M. Gallagher
  Expert Review of Molecular Diagnostics.2020; 20(10): 1027.     CrossRef
• Current Trends of Artificial Intelligence for Colorectal Cancer Pathology Image Analysis: A Systematic Review
  Nishant Thakur, Hongjun Yoon, Yosep Chong
  Cancers.2020; 12(7): 1884.     CrossRef
• Explainable Machine Learning Model for Predicting GI Bleed Mortality in the Intensive Care Unit
  Farah Deshmukh, Shamel S. Merchant
  American Journal of Gastroenterology.2020; 115(10): 1657.     CrossRef
• Prediction of clinically actionable genetic alterations from colorectal cancer histopathology images using deep learning
  Hyun-Jong Jang, Ahwon Lee, J Kang, In Hye Song, Sung Hak Lee
  World Journal of Gastroenterology.2020; 26(40): 6207.     CrossRef
• Application of system analysis methods for modeling the development of hand-arm vibration syndrome: problems and approaches to solution
  M P Diakovich, M V Krivov
  Journal of Physics: Conference Series.2020; 1661(1): 012029.     CrossRef
• Histo-ELISA technique for quantification and localization of tissue components
  Zhongmin Li, Silvia Goebel, Andreas Reimann, Martin Ungerer
  Scientific Reports.2020;[Epub]     CrossRef
• Role of artificial intelligence in diagnostic oral pathology-A modern approach
  AyinampudiBhargavi Krishna, Azra Tanveer, PanchaVenkat Bhagirath, Ashalata Gannepalli
  Journal of Oral and Maxillofacial Pathology.2020; 24(1): 152.     CrossRef
• Applications of deep learning for the analysis of medical data
  Hyun-Jong Jang, Kyung-Ok Cho
  Archives of Pharmacal Research.2019; 42(6): 492.     CrossRef
• PROMISE CLIP Project: A Retrospective, Multicenter Study for Prostate Cancer that Integrates Clinical, Imaging and Pathology Data
  Jihwan Park, Mi Jung Rho, Yong Hyun Park, Chan Kwon Jung, Yosep Chong, Choung-Soo Kim, Heounjeong Go, Seong Soo Jeon, Minyong Kang, Hak Jong Lee, Sung Il Hwang, Ji Youl Lee
  Applied Sciences.2019; 9(15): 2982.     CrossRef
• Key challenges for delivering clinical impact with artificial intelligence
  Christopher J. Kelly, Alan Karthikesalingam, Mustafa Suleyman, Greg Corrado, Dominic King
  BMC Medicine.2019;[Epub]     CrossRef
• Deep Learning for Whole Slide Image Analysis: An Overview
  Neofytos Dimitriou, Ognjen Arandjelović, Peter D. Caie
  Frontiers in Medicine.2019;[Epub]     CrossRef
• Barriers to Artificial Intelligence Adoption in Healthcare Management: A Systematic Review
  Mir Mohammed Assadullah
  SSRN Electronic Journal .2019;[Epub]     CrossRef