Assessment of deep learning assistance for the pathological diagnosis of gastric cancer.

Previous studies on deep learning (DL) applications in pathology have focused on pathologist-versus-algorithm comparisons. However, DL will not replace the breadth and contextual knowledge of pathologists; rather, only through their combination may the benefits of DL be achieved. A fully crossed multireader multicase study was conducted to evaluate DL assistance with pathologists' diagnosis of gastric cancer. A total of 110 whole-slide images (WSI) (50 malignant and 60 benign) were interpreted by 16 board-certified pathologists with or without DL assistance, with a washout period between sessions. DL-assisted pathologists achieved a higher area under receiver operating characteristic curve (ROC-AUC) (0.911 vs. 0.863, P = 0.003) than unassisted in interpreting the 110 WSIs. Pathologists with DL assistance demonstrated higher sensitivity in detection of gastric cancer than without (90.63% vs. 82.75%, P = 0.010). No significant difference was observed in specificity with or without deep learning assistance (78.23% vs. 79.90%, P = 0.468). The average review time per WSI was shortened with DL assistance than without (22.68 vs. 26.37 second, P = 0.033). Our results demonstrated that DL assistance indeed improved pathologists' accuracy and efficiency in gastric cancer diagnosis and further boosted the acceptance of this new technique.

Histopathological Diagnosis System for Gastritis Using Deep Learning Algorithm.

Objective To develope a deep learning algorithm for pathological classification of chronic gastritis and assess its performance using whole-slide images (WSIs). Methods We retrospectively collected 1,250 gastric biopsy specimens (1,128 gastritis, 122 normal mucosa) from PLA General Hospital. The deep learning algorithm based on DeepLab v3 (ResNet-50) architecture was trained and validated using 1,008 WSIs and 100 WSIs, respectively. The diagnostic performance of the algorithm was tested on an independent test set of 142 WSIs, with the pathologists' consensus diagnosis as the gold standard. Results The receiver operating characteristic (ROC) curves were generated for chronic superficial gastritis (CSuG), chronic active gastritis (CAcG), and chronic atrophic gastritis (CAtG) in the test set, respectively.The areas under the ROC curves (AUCs) of the algorithm for CSuG, CAcG, and CAtG were 0.882, 0.905 and 0.910, respectively. The sensitivity and specificity of the deep learning algorithm for the classification of CSuG, CAcG, and CAtG were 0.790 and 1.000 (accuracy 0.880), 0.985 and 0.829 (accuracy 0.901), 0.952 and 0.992 (accuracy 0.986), respectively. The overall predicted accuracy for three different types of gastritis was 0.867. By flagging the suspicious regions identified by the algorithm in WSI, a more transparent and interpretable diagnosis can be generated. Conclusion The deep learning algorithm achieved high accuracy for chronic gastritis classification using WSIs. By pre-highlighting the different gastritis regions, it might be used as an auxiliary diagnostic tool to improve the work efficiency of pathologists.

Automatic detection of squamous cell carcinoma metastasis in esophageal lymph nodes using semantic segmentation.

Esophageal squamous cell carcinoma (ESCC) is more prevalent than esophageal adenocarcinoma in Asia, especially in China, where more than half of ESCC cases occur worldwide. Many studies have reported that the automatic detection of lymph node metastasis using semantic segmentation shows good performance in breast cancer and other adenocarcinomas. However, the detection of squamous cell carcinoma metastasis in hematoxylin-eosin (H&E)-stained slides has never been reported. We collected a training set of 110 esophageal lymph node slides with metastasis and 132 lymph node slides without metastasis. An iPad-based annotation system was used to draw the contours of the cancer metastasis region. A DeepLab v3 model was trained to achieve the best fit with the training data. The learned model could estimate the probability of metastasis. To evaluate the effectiveness of the detection model of learned metastasis, we used another large cohort of clinical H&E-stained esophageal lymph node slides containing 795 esophageal lymph nodes from 154 esophageal cancer patients. The basic authenticity label for each slide was confirmed by experienced pathologists. After filtering isolated noise in the prediction, we obtained an accuracy of 94%. Furthermore, we applied the learned model to throat and lung lymph node squamous cell carcinoma metastases and achieved the following promising results: an accuracy of 96.7% in throat cancer and an accuracy of 90% in lung cancer. In this work, we organized an annotated dataset of H&E-stained esophageal lymph node and trained a deep neural network to detect lymph node metastasis in H&E-stained slides of squamous cell carcinoma automatically. Moreover, it is possible to use this model to detect lymph nodes metastasis in squamous cell carcinoma from other organs. This study directly demonstrates the potential for determining the localization of squamous cell carcinoma metastases in lymph node and assisting in pathological diagnosis.

Clinically applicable histopathological diagnosis system for gastric cancer detection using deep learning.

The early detection and accurate histopathological diagnosis of gastric cancer increase the chances of successful treatment. The worldwide shortage of pathologists offers a unique opportunity for the use of artificial intelligence assistance systems to alleviate the workload and increase diagnostic accuracy. Here, we report a clinically applicable system developed at the Chinese PLA General Hospital, China, using a deep convolutional neural network trained with 2,123 pixel-level annotated H&E-stained whole slide images. The model achieves a sensitivity near 100% and an average specificity of 80.6% on a real-world test dataset with 3,212 whole slide images digitalized by three scanners. We show that the system could aid pathologists in improving diagnostic accuracy and preventing misdiagnoses. Moreover, we demonstrate that our system performs robustly with 1,582 whole slide images from two other medical centres. Our study suggests the feasibility and benefits of using histopathological artificial intelligence assistance systems in routine practice scenarios.

Automatic deep learning-based colorectal adenoma detection system and its similarities with pathologists.

OBJECTIVES: The microscopic evaluation of slides has been gradually moving towards all digital in recent years, leading to the possibility for computer-aided diagnosis. It is worthwhile to know the similarities between deep learning models and pathologists before we put them into practical scenarios. The simple criteria of colorectal adenoma diagnosis make it to be a perfect testbed for this study. DESIGN: The deep learning model was trained by 177 accurately labelled training slides (156 with adenoma). The detailed labelling was performed on a self-developed annotation system based on iPad. We built the model based on DeepLab v2 with ResNet-34. The model performance was tested on 194 test slides and compared with five pathologists. Furthermore, the generalisation ability of the learning model was tested by extra 168 slides (111 with adenoma) collected from two other hospitals. RESULTS: The deep learning model achieved an area under the curve of 0.92 and obtained a slide-level accuracy of over 90% on slides from two other hospitals. The performance was on par with the performance of experienced pathologists, exceeding the average pathologist. By investigating the feature maps and cases misdiagnosed by the model, we found the concordance of thinking process in diagnosis between the deep learning model and pathologists. CONCLUSIONS: The deep learning model for colorectal adenoma diagnosis is quite similar to pathologists. It is on-par with pathologists' performance, makes similar mistakes and learns rational reasoning logics. Meanwhile, it obtains high accuracy on slides collected from different hospitals with significant staining configuration variations.