Artificial Intelligence-Assisted Image Analysis of Acetaminophen-Induced Acute Hepatic Injury in Sprague-Dawley Rats

Abstract

Although drug-induced liver injury (DILI) is a major target of the pharmaceutical industry, we currently lack an efficient model for evaluating liver toxicity in the early stage of its development. Recent progress in artificial intelligence-based deep learning technology promises to improve the accuracy and robustness of current toxicity prediction models. Mask region-based CNN (Mask R-CNN) is a detection-based segmentation model that has been used for developing algorithms. In the present study, we applied a Mask R-CNN algorithm to detect and predict acute hepatic injury lesions induced by acetaminophen (APAP) in Sprague-Dawley rats. To accomplish this, we trained, validated, and tested the model for various hepatic lesions, including necrosis, inflammation, infiltration, and portal triad. We confirmed the model performance at the whole-slide image (WSI) level. The training, validating, and testing processes, which were performed using tile images, yielded an overall model accuracy of 96.44%. For confirmation, we compared the model’s predictions for 25 WSIs at 20× magnification with annotated lesion areas determined by an accredited toxicologic pathologist. In individual WSIs, the expert-annotated lesion areas of necrosis, inflammation, and infiltration tended to be comparable with the values predicted by the algorithm. The overall predictions showed a high correlation with the annotated area. The R square values were 0.9953, 0.9610, and 0.9445 for necrosis, inflammation plus infiltration, and portal triad, respectively. The present study shows that the Mask R-CNN algorithm is a useful tool for detecting and predicting hepatic lesions in non-clinical studies. This new algorithm might be widely useful for predicting liver lesions in non-clinical and clinical settings.

A comparative study on the implementation of deep learning algorithms for detection of hepatic necrosis in toxicity studies

“…The accuracy values in this study are lower than those of previous research due to the false detection of other features such as inflammatory cells and RBCs as hepatic necrosis. The incorrect prediction observed in this study could be resolved by annotating cells that are often mistaken for lesions and including their exclusion in the training, together with the hepatic necrosis, as shown in our previous study [ 17 ].…”

“…We suggest that this model is relatively reliable for detecting a single lesion of interest. Previous studies have shown nearly 100% accuracy in detecting hepatic necrosis when using a consolidated model trained with various other lesions [ 3 , 17 ]. The accuracy values in this study are lower than those of previous research due to the false detection of other features such as inflammatory cells and RBCs as hepatic necrosis.…”

“…To induce hepatic necrosis in Sprague–Dawley (SD) rats, we conducted acute oral toxicity tests as explained previously [ 17 ]. Male and female SD rats (Crl:CD) were obtained from Orient Bio, Inc. (Gyeonggi, Korea) at 9 weeks old.…”

“…Segmentation architectures include fully convolution networks [ 12 ], U-NET [ 13 ], DeepLab [ 14 ], and Mask region-based CNN (Mask R-CNN) [ 15 ]. Our previous studies have shown the successful implementation of Mask R-CNN to detect and quantify the degree of hepatic fibrosis [ 16 ] and hepatic lesions involved in acute hepatic injury [ 17 ] at the WSI level. However, the evaluation and comparative examination of Mask R-CNN and other deep learning algorithms are necessary for determining an optimal model for detecting lesions in non-clinical studies.…”

Application of multiple-finding segmentation utilizing Mask R-CNN-based deep learning in a rat model of drug-induced liver injury

“…Medical imaging research has explored various machine-learning techniques, including many classifier and clustering algorithms 15 . A technique that has shown great promise is the mask region-based convolutional neural network (Mask R-CNN), a detection-based segmentation model 16 , 17 comprised of two main stages: (i) object detection and localization; and (ii) using the features of the detected regions to classify them, assign their final localization, and segment them 16 . Recently, Mask R-CNN-based approaches have been used in medical research 17 – 19 .…”