Preparing pathological data to develop an artificial intelligence model in the nonclinical study

Abstract

Artificial intelligence (AI)-based analysis has recently been adopted in the examination of histological slides via the digitization of glass slides using a digital scanner. In this study, we examined the effect of varying the staining color tone and magnification level of a dataset on the result of AI model prediction in hematoxylin and eosin stained whole slide images (WSIs). The WSIs of liver tissues with fibrosis were used as an example, and three different datasets (N20, B20, and B10) were prepared with different color tones and magnifications. Using these datasets, we built five models trained Mask R-CNN algorithm by a single or mixed dataset of N20, B20, and B10. We evaluated their model performance using the test dataset of three datasets. It was found that the models that were trained with mixed datasets (models B20/N20 and B10/B20), which consist of different color tones or magnifications, performed better than the single dataset trained models. Consequently, superior performance of the mixed models was obtained from the actual prediction results of the test images. We suggest that training the algorithm with various staining color tones and multi-scaled image datasets would be more optimized for consistent remarkable performance in predicting pathological lesions of interest.

The Complexity of Shapes: How the Circularity of Tumor Nodules Affects Prognosis in Colorectal Cancer

“…This made the cutoff for maximum complexity suboptimal for this cohort. The issues that come with interinstitute variation regarding the quality of stains are wellknown in histopathologic artificial intelligence (AI), 29 and these can only be resolved by further innovations in this field such as color-optimizing algorithms. For the severe discoloration in our validation cohort, no suitable solution was available.…”

The Use of Artificial Intelligence in the Liver Histopathology Field: A Systematic Review

“…Recently, CV has shown promise in supporting the diagnosis and treatment of liver disease [54]. CV algorithms can be used to detect and segment liver lesions, such as tumors or cysts [55,56], to analyze liver texture and structure to identify areas of fibrosis or cirrhosis [57][58][59], and to develop models that can predict liver disease progression [60][61][62][63][64][65][66].…”

“…An automated tool for detecting and quantifying liver fibrosis on digital images of trichrome-stained slides of patients with NAFLD was proposed by Gawrieh et al [57]. The detection of different types of fibrosis was performed by SVM classifiers with linear kernels trained using morphological features and structural properties of the blue areas extracted from pathologist annotation and correlated with the presence of collagen.…”