Automated analysis of small intestinal lamina propria to distinguish normal, Celiac Disease, and Non-Celiac Duodenitis biopsy images

BACKGROUND AND OBJECTIVE

Celiac Disease (CD) is characterized by gluten intolerance in genetically predisposed individuals. High disease prevalence, absence of a cure, and low diagnosis rates make this disease a public health problem. The diagnosis of CD predominantly relies on recognizing characteristic mucosal alterations of the small intestine, such as villous atrophy, crypt hyperplasia, and intraepithelial lymphocytosis. However, these changes are not entirely specific to CD and overlap with Non-Celiac Duodenitis (NCD) due to various etiologies. We investigated whether Artificial Intelligence (AI) models could assist in distinguishing normal, CD, and NCD (and unaffected individuals) based on the characteristics of small intestinal lamina propria (LP).

METHODS

Our method was developed using a dataset comprising high magnification biopsy images of the duodenal LP compartment of CD patients with different clinical stages of CD, those with NCD, and individuals lacking an intestinal inflammatory disorder (controls). A pre-processing step was used to standardize and enhance the acquired images.

RESULTS

For the normal controls versus CD use case, a Support Vector Machine (SVM) achieved an Accuracy (ACC) of 98.53%. For a second use case, we investigated the ability of the classification algorithm to differentiate between normal controls and NCD. In this use case, the SVM algorithm with linear kernel outperformed all the tested classifiers by achieving 98.55% ACC.

CONCLUSIONS

To the best of our knowledge, this is the first study that documents automated differentiation between normal, NCD, and CD biopsy images. These findings are a stepping stone toward automated biopsy image analysis that can significantly benefit patients and healthcare providers.

Human Polyomavirus 9-An Emerging Cutaneous and Pulmonary Pathogen in Solid Organ Transplant Recipients

IMPORTANCE

We describe the first report to our knowledge of cutaneous and systemic pathogenicity of human polyomavirus 9 in solid organ transplant recipients.

OBJECTIVE

Three solid organ transplant recipients developed a widespread, progressive, violaceous, and hyperkeratotic skin eruption. All died from pulmonary and multiorgan failure around 1 year from onset of the rash. Routine clinical diagnostic testing could not identify any causative agent; therefore, samples and autopsies were investigated for novel pathogens using high-throughput sequencing.

DESIGN, SETTING, AND PARTICIPANTS

This case series, including 3 solid organ transplant recipients who developed characteristic pink, violaceous, or brown hyperkeratotic papules and plaques throughout the body, was conducted at the Columbia University Medical Center. Lesional skin biopsies were collected from all 3 patients and subjected to high-throughput illumina sequencing for identification of microbial pathogens. Human polyomavirus 9 was identified in lesional skin biopsies. We subsequently collected ocular swabs, oral swabs, urine samples, and blood samples from patients, and organ tissues at autopsy in 1 patient. We investigated these samples for the presence of human polyomavirus 9 using in situ hybridization and quantitative polymerase chain reaction (PCR) assays.

MAIN OUTCOMES AND MEASURES

A description of the clinical and pathologic findings of 3 patients.

RESULTS

This case series study found that human polyomavirus 9 was detected in the skin biopsies of all 3 patients by a capture-based high-throughput sequencing method platform (VirCapSeq-VERT). Human polyomavirus 9 was also detected in blood, oral, ocular swabs, and urine by real-time polymerase chain reaction (PCR) assay. In situ hybridization and quantitative PCR assays were performed on the skin biopsies from 3 patients and lung autopsy of 1 patient, which showed the presence of human polyomavirus 9 messenger RNA transcripts, indicating active viral replication and pathogenesis in the skin and lungs.

CONCLUSIONS AND RELEVANCE

Human polyomavirus 9 was associated with the widespread cutaneous eruption. All 3 patients had progression of cutaneous disease, accompanied by clinical deterioration, pulmonary failure, and death. One patient underwent autopsy and human polyomavirus 9 was identified in the lungs and paratracheal soft tissue. These findings suggest that human polyomavirus 9 may be associated with cutaneous and possibly pulmonary infection and death in solid organ transplant recipients.

Automated interpretation of biopsy images for the detection of celiac disease using a machine learning approach

BACKGROUND AND OBJECTIVES

Celiac disease is an autoimmune disease occurring in about 1 in 100 people worldwide. Early diagnosis and efficient treatment are crucial in mitigating the complications that are associated with untreated celiac disease, such as intestinal lymphoma and malignancy, and the subsequent high morbidity. The current diagnostic methods using small intestinal biopsy histopathology, endoscopy, and video capsule endoscopy (VCE) involve manual interpretation of photomicrographs or images, which can be time-consuming and difficult, with inter-observer variability. In this paper, a machine learning technique was developed for the automation of biopsy image analysis to detect and classify villous atrophy based on modified Marsh scores. This is one of the first studies to employ conventional machine learning to automate the use of biopsy images for celiac disease detection and classification.

METHODS

The Steerable Pyramid Transform (SPT) method was used to obtain sub bands from which various types of entropy and nonlinear features were computed. All extracted features were automatically classified into two-class and multi-class, using six classifiers.

RESULTS

An accuracy of 88.89%, was achieved for the classification of two-class villous abnormalities based on analysis of Hematoxylin and Eosin (H&E) stained biopsy images. Similarly, an accuracy of 82.92% was achieved for the two-class classification of red-green-blue (RGB) biopsy images. Also, an accuracy of 72% was achieved in the classification of multi-class biopsy images.

CONCLUSION

The results obtained are promising, and demonstrate the possibility of automating biopsy image interpretation using machine learning. This can assist pathologists in accelerating the diagnostic process without bias, resulting in greater accuracy, and ultimately, earlier access to treatment.

SATB2 in Neoplasms of Lung, Pancreatobiliary, and Gastrointestinal Origins

OBJECTIVES

Special AT-rich binding protein 2 (SATB2) immunohistochemistry (IHC) has high sensitivity and specificity for colorectal adenocarcinoma (CRC), but data on its expression in specific subsets of pulmonary, gastric, small bowel, and pancreatobiliary adenocarcinomas (ADCAs) are relatively limited or discordant. We assessed SATB2 expression in a large cohort of ADCAs from these sites to determine its reliability in distinguishing CRC from them.

METHODS

SATB2 IHC was performed on 335 neoplasms, including 40 lung ADCAs, 165 pancreatobiliary neoplasms (34 intraductal papillary mucinous neoplasms [IPMNs], 19 pancreatic ADCAs, 112 cholangiocarcinomas [CCs]), and 35 gastric, 13 small bowel, 36 ampullary (AMP), and 46 CRC ADCAs. The cases were evaluated for positivity (defined as ≥5% nuclear staining), and an H-score was calculated based on the percentage of SATB2+ cells and staining intensity. Analysis was performed to determine the optimal H-score threshold to separate CRC and non-CRC.

RESULTS

SATB2 was positive in 3% of lung, 2% of CC, 17% of gastric, 38% of small bowel, and 6% of AMP ADCAs. All pancreatic ADCA/IPMNs were negative, and 87% CRCs were positive.

CONCLUSIONS

SATB2 is not entirely specific for colorectal origin and can be expressed in a subset of gastrointestinal ADCAs. It is most useful in the differential of CRC vs lung and pancreatobiliary ADCAs.

Forty Postmortem Examinations in COVID-19 Patients

OBJECTIVES

Although diffuse alveolar damage, a subtype of acute lung injury (ALI), is the most common microscopic pattern in coronavirus disease 2019 (COVID-19), other pathologic patterns have been described. The aim of the study was to review autopsies from COVID-19 decedents to evaluate the spectrum of pathology and correlate the results with clinical, laboratory, and radiologic findings.

METHODS

A comprehensive and quantitative review from 40 postmortem examinations was performed. The microscopic patterns were categorized as follows: "major" when present in more than 50% of cases and "novel" if rarely or not previously described and unexpected clinically.

RESULTS

Three major pulmonary patterns were identified: ALI in 29 (73%) of 40, intravascular fibrin or platelet-rich aggregates (IFPAs) in 36 (90%) of 40, and vascular congestion and hemangiomatosis-like change (VCHL) in 20 (50%) of 40. The absence of ALI (non-ALI) was novel and seen in 11 (27%) of 40. Compared with ALI decedents, those with non-ALI had a shorter hospitalization course (P = .02), chest radiographs with no or minimal consolidation (P = .01), and no pathologically confirmed cause of death (9/11). All non-ALI had VCHL and IFPAs, and clinically most had cardiac arrest.

CONCLUSIONS

Two distinct pulmonary phenotypic patterns-ALI and non-ALI-were noted. Non-ALI represents a rarely described phenotype. The cause of death in non-ALI is most likely COVID-19 related but requires additional corroboration.

COVID-19 Associated Hepatitis Complicating Recent Living Donor Liver Transplantation

We present a case of COVID-19 hepatitis in a living donor liver allograft recipient whose donor subsequently tested positive for COVID-19. The patient is a female infant with biliary atresia (failed Kasai procedure). She recovered well, with improving liver function tests for 4 days. On post-operative day (POD) 4 the patient developed respiratory distress and fever. COVID-19 testing (polymerase chain reaction) was positive. Liver function tests increased approximately 5-fold. Liver biopsy showed moderate acute hepatitis with prominent clusters of apoptotic hepatocytes and associated cellular debris. Lobular lymphohistiocytic inflammation was noted. Typical portal features of mild to moderate acute cellular rejection were also noted.