Background and Aims Nonalcoholic steatohepatitis (NASH) is a common cause of chronic liver disease. Clinical trials use the NASH Clinical Research Network (CRN) system for semiquantitative histological assessment of disease severity. Interobserver variability may hamper histological assessment, and diagnostic consensus is not always achieved. We evaluate a second harmonic generation/two‐photon excitation fluorescence (SHG/TPEF) imaging‐based tool to provide an automated quantitative assessment of histological features pertinent to NASH. Approach and Results Images were acquired by SHG/TPEF from 219 nonalcoholic fatty liver disease (NAFLD)/NASH liver biopsy samples from seven centers in Asia and Europe. These were used to develop and validate qFIBS, a computational algorithm that quantifies key histological features of NASH. qFIBS was developed based on in silico analysis of selected signature parameters for four cardinal histopathological features, that is, fibrosis (qFibrosis), inflammation (qInflammation), hepatocyte ballooning (qBallooning), and steatosis (qSteatosis), treating each as a continuous rather than categorical variable. Automated qFIBS analysis outputs showed strong correlation with each respective component of the NASH CRN scoring (P < 0.001; qFibrosis [r = 0.776], qInflammation [r = 0.557], qBallooning [r = 0.533], and qSteatosis [r = 0.802]) and high area under the receiver operating characteristic curve values (qFibrosis [0.870‐0.951; 95% confidence interval {CI}, 0.787‐1.000; P < 0.001], qInflammation [0.820‐0.838; 95% CI, 0.726‐0.933; P < 0.001), qBallooning [0.813‐0.844; 95% CI, 0.708‐0.957; P < 0.001], and qSteatosis [0.939‐0.986; 95% CI, 0.867‐1.000; P < 0.001]) and was able to distinguish differing grades/stages of histological disease. Performance of qFIBS was best when assessing degree of steatosis and fibrosis, but performed less well when distinguishing severe inflammation and higher ballooning grades. Conclusions qFIBS is an automated tool that accurately quantifies the critical components of NASH histological assessment. It offers a tool that could potentially aid reproducibility and standardization of liver biopsy assessments required for NASH therapeutic clinical trials.

Colorectal cancer is one of the most common cancers worldwide, accounting for an annual estimated 1.8 million incident cases. With the increasing number of colonoscopies being performed, colorectal biopsies make up a large proportion of any histopathology laboratory workload. We trained and validated a unique artificial intelligence (AI) deep learning model as an assistive tool to screen for colonic malignancies in colorectal specimens, in order to improve cancer detection and classification; enabling busy pathologists to focus on higher order decision-making tasks. The study cohort consists of Whole Slide Images (WSI) obtained from 294 colorectal specimens. Qritive’s unique composite algorithm comprises both a deep learning model based on a Faster Region Based Convolutional Neural Network (Faster-RCNN) architecture for instance segmentation with a ResNet-101 feature extraction backbone that provides glandular segmentation, and a classical machine learning classifier. The initial training used pathologists’ annotations on a cohort of 66,191 image tiles extracted from 39 WSIs. A subsequent application of a classical machine learning-based slide classifier sorted the WSIs into ‘low risk’ (benign, inflammation) and ‘high risk’ (dysplasia, malignancy) categories. We further trained the composite AI-model’s performance on a larger cohort of 105 resections WSIs and then validated our findings on a cohort of 150 biopsies WSIs against the classifications of two independently blinded pathologists. We evaluated the area under the receiver-operator characteristic curve (AUC) and other performance metrics. The AI model achieved an AUC of 0.917 in the validation cohort, with excellent sensitivity (97.4%) in detection of high risk features of dysplasia and malignancy. We demonstrate an unique composite AI-model incorporating both a glandular segmentation deep learning model and a classical machine learning classifier, with excellent sensitivity in picking up high risk colorectal features. As such, AI plays a role as a potential screening tool in assisting busy pathologists by outlining the dysplastic and malignant glands.

Background Conventional differential expression (DE) testing compares the grouped mean value of tumour samples to the grouped mean value of the normal samples, and may miss out dysregulated genes in small subgroup of patients. This is especially so for highly heterogeneous cancer like Hepatocellular Carcinoma (HCC). Methods Using multi-region sampled RNA-seq data of 90 patients, we performed patient-specific differential expression testing, together with the patients’ matched adjacent normal samples. Results Comparing the results from conventional DE analysis and patient-specific DE analyses, we show that the conventional DE analysis omits some genes due to high inter-individual variability present in both tumour and normal tissues. Dysregulated genes shared in small subgroup of patients were useful in stratifying patients, and presented differential prognosis. We also showed that the target genes of some of the current targeted agents used in HCC exhibited highly individualistic dysregulation pattern, which may explain the poor response rate. Discussion/conclusion Our results highlight the importance of identifying patient-specific DE genes, with its potential to provide clinically valuable insights into patient subgroups for applications in precision medicine.