Machine-Learning-Based Classification Model to Address Diagnostic Challenges in Transbronchial Lung Biopsy.

BACKGROUND: When obtaining specimens from pulmonary nodules in TBLB, distinguishing between benign samples and mis-sampling from a tumor presents a challenge. Our objective is to develop a machine-learning-based classifier for TBLB specimens. METHODS: Three pathologists assessed six pathological findings, including interface bronchitis/bronchiolitis (IB/B), plasma cell infiltration (PLC), eosinophil infiltration (Eo), lymphoid aggregation (Ly), fibroelastosis (FE), and organizing pneumonia (OP), as potential histologic markers to distinguish between benign and malignant conditions. A total of 251 TBLB cases with defined benign and malignant outcomes based on clinical follow-up were collected and a gradient-boosted decision-tree-based machine learning model (XGBoost) was trained and tested on randomly split training and test sets. RESULTS: Five pathological changes showed independent, mild-to-moderate associations (AUC ranging from 0.58 to 0.75) with benign conditions, with IB/B being the strongest predictor. On the other hand, FE emerged to be the sole indicator of malignant conditions with a mild association (AUC = 0.66). Our model was trained on 200 cases and tested on 51 cases, achieving an AUC of 0.78 for the binary classification of benign vs. malignant on the test set. CONCLUSION: The machine-learning model developed has the potential to distinguish between benign and malignant conditions in TBLB samples excluding the presence or absence of tumor cells, thereby improving diagnostic accuracy and reducing the burden of repeated sampling procedures for patients.

Standardized Classification of Lung Adenocarcinoma Subtypes and Improvement of Grading Assessment Through Deep Learning.

The histopathologic distinction of lung adenocarcinoma (LADC) subtypes is subject to high interobserver variability, which can compromise the optimal assessment of patient prognosis. Therefore, this study developed convolutional neural networks capable of distinguishing LADC subtypes and predicting disease-specific survival, according to the recently established LADC tumor grades. Consensus LADC histopathologic images were obtained from 17 expert pulmonary pathologists and one pathologist in training. Two deep learning models (AI-1 and AI-2) were trained to predict eight different LADC classes. Furthermore, the trained models were tested on an independent cohort of 133 patients. The models achieved high precision, recall, and F1 scores exceeding 0.90 for most of the LADC classes. Clear stratification of the three LADC grades was reached in predicting the disease-specific survival by the two models, with both Kaplan-Meier curves showing significance (P = 0.0017 and 0.0003). Moreover, both trained models showed high stability in the segmentation of each pair of predicted grades with low variation in the hazard ratio across 200 bootstrapped samples. These findings indicate that the trained convolutional neural networks improve the diagnostic accuracy of the pathologist and refine LADC grade assessment. Thus, the trained models are promising tools that may assist in the routine evaluation of LADC subtypes and grades in clinical practice.

Effect of the 2020 hypersensitivity pneumonitis guideline on the pathologic diagnosis of interstitial pneumonia.

It was reported that the 2020 guideline for hypersensitivity pneumonitis (HP) might result in the overdiagnosis of fibrotic HP (fHP). fHP and other types of interstitial pneumonias have several overlapping characteristics, and a high diagnostic concordance rate of fHP is rarely obtained. Therefore, we investigated the impact of the 2020 HP guideline on the pathological diagnosis of cases previously diagnosed as interstitial pneumonia. We identified 289 fibrotic interstitial pneumonia cases from 2014 to 2019 and classified them into four categories according to the 2020 HP guideline: typical, probable, and indeterminate for fHP and alternative diagnosis. The original pathological diagnosis of 217 cases were compared to their classification as either typical, probable, or indeterminate for fHP according to the 2020 guideline. The clinical data, including serum data and pulmonary function tests, were compared among the groups. Diagnoses changed from non-fHP to fHP for 54 (25%) of the 217 cases, of which, 8 were typical fHP and 46 were probable fHP. The ratio of typical and probable fHP cases to the total number of VATS cases was significantly lower when using transbronchial lung cryobiopsy (p < 0.001). The clinical data of these cases bore a more remarkable resemblance to those diagnosed as indeterminate for fHP than to those diagnosed as typical or probable. The pathological criteria in the new HP guidelines increase the diagnosis of fHP. However, it is unclear whether this increase leads to overdiagnosis, and requires further investigation. Transbronchial lung cryobiopsy may not be helpful when using the new criteria to impart findings for fHP diagnosis.

Generalizable biomarker prediction from cancer pathology slides with self-supervised deep learning: A retrospective multi-centric study.

Deep learning (DL) can predict microsatellite instability (MSI) from routine histopathology slides of colorectal cancer (CRC). However, it is unclear whether DL can also predict other biomarkers with high performance and whether DL predictions generalize to external patient populations. Here, we acquire CRC tissue samples from two large multi-centric studies. We systematically compare six different state-of-the-art DL architectures to predict biomarkers from pathology slides, including MSI and mutations in BRAF, KRAS, NRAS, and PIK3CA. Using a large external validation cohort to provide a realistic evaluation setting, we show that models using self-supervised, attention-based multiple-instance learning consistently outperform previous approaches while offering explainable visualizations of the indicative regions and morphologies. While the prediction of MSI and BRAF mutations reaches a clinical-grade performance, mutation prediction of PIK3CA, KRAS, and NRAS was clinically insufficient.

MIXTURE of human expertise and deep learning-developing an explainable model for predicting pathological diagnosis and survival in patients with interstitial lung disease.

Interstitial pneumonia is a heterogeneous disease with a progressive course and poor prognosis, at times even worse than those in the main cancer types. Histopathological examination is crucial for its diagnosis and estimation of prognosis. However, the evaluation strongly depends on the experience of pathologists, and the reproducibility of diagnosis is low. Herein, we propose MIXTURE (huMan-In-the-loop eXplainable artificial intelligence Through the Use of REcurrent training), an original method to develop deep learning models for extracting pathologically significant findings based on an expert pathologist's perspective with a small annotation effort. The procedure of MIXTURE consists of three steps as follows. First, we created feature extractors for tiles from whole slide images using self-supervised learning. The similar looking tiles were clustered based on the output features and then pathologists integrated the pathologically synonymous clusters. Using the integrated clusters as labeled data, deep learning models to classify the tiles into pathological findings were created by transfer-learning the feature extractors. We developed three models for different magnifications. Using these extracted findings, our model was able to predict the diagnosis of usual interstitial pneumonia, a finding suggestive of progressive disease, with high accuracy (AUC 0.90 in validation set and AUC 0.86 in test set). This high accuracy could not be achieved without the integration of findings by pathologists. The patients predicted as UIP had poorer prognosis (5-year overall survival [OS]: 55.4%) than those predicted as non-UIP (OS: 95.2%). The Cox proportional hazards model for each microscopic finding and prognosis pointed out dense fibrosis, fibroblastic foci, elastosis, and lymphocyte aggregation as independent risk factors. We suggest that MIXTURE may serve as a model approach to different diseases evaluated by medical imaging, including pathology and radiology, and be the prototype for explainable artificial intelligence that can collaborate with humans.

A narrative review of digital pathology and artificial intelligence: focusing on lung cancer.

The emergence of whole slide imaging technology allows for pathology diagnosis on a computer screen. The applications of digital pathology are expanding, from supporting remote institutes suffering from a shortage of pathologists to routine use in daily diagnosis including that of lung cancer. Through practice and research large archival databases of digital pathology images have been developed that will facilitate the development of artificial intelligence (AI) methods for image analysis. Currently, several AI applications have been reported in the field of lung cancer; these include the segmentation of carcinoma foci, detection of lymph node metastasis, counting of tumor cells, and prediction of gene mutations. Although the integration of AI algorithms into clinical practice remains a significant challenge, we have implemented tumor cell count for genetic analysis, a helpful application for routine use. Our experience suggests that pathologists often overestimate the contents of tumor cells, and the use of AI-based analysis increases the accuracy and makes the tasks less tedious. However, there are several difficulties encountered in the practical use of AI in clinical diagnosis. These include the lack of sufficient annotated data for the development and validation of AI systems, the explainability of black box AI models, such as those based on deep learning that offer the most promising performance, and the difficulty in defining the ground truth data for training and validation owing to inherent ambiguity in most applications. All of these together present significant challenges in the development and clinical translation of AI methods in the practice of pathology. Additional research on these problems will help in resolving the barriers to the clinical use of AI. Helping pathologists in developing knowledge of the working and limitations of AI will benefit the use of AI in both diagnostics and research.

Value of laboratory results in addition to vital signs in a machine learning algorithm to predict in-hospital cardiac arrest: A single-center retrospective cohort study.

BACKGROUND: Although machine learning-based prediction models for in-hospital cardiac arrest (IHCA) have been widely investigated, it is unknown whether a model based on vital signs alone (Vitals-Only model) can perform similarly to a model that considers both vital signs and laboratory results (Vitals+Labs model). METHODS: All adult patients hospitalized in a tertiary care hospital in Japan between October 2011 and October 2018 were included in this study. Random forest models with/without laboratory results (Vitals+Labs model and Vitals-Only model, respectively) were trained and tested using chronologically divided datasets. Both models use patient demographics and eight-hourly vital signs collected within the previous 48 hours. The primary and secondary outcomes were the occurrence of IHCA in the next 8 and 24 hours, respectively. The area under the receiver operating characteristic curve (AUC) was used as a comparative measure. Sensitivity analyses were performed under multiple statistical assumptions. RESULTS: Of 141,111 admitted patients (training data: 83,064, test data: 58,047), 338 had an IHCA (training data: 217, test data: 121) during the study period. The Vitals-Only model and Vitals+Labs model performed comparably when predicting IHCA within the next 8 hours (Vitals-Only model vs Vitals+Labs model, AUC = 0.862 [95% confidence interval (CI): 0.855-0.868] vs 0.872 [95% CI: 0.867-0.878]) and 24 hours (Vitals-Only model vs Vitals+Labs model, AUC = 0.830 [95% CI: 0.825-0.835] vs 0.837 [95% CI: 0.830-0.844]). Both models performed similarly well on medical, surgical, and ward patient data, but did not perform well for intensive care unit patients. CONCLUSIONS: In this single-center study, the machine learning model predicted IHCAs with good discrimination. The addition of laboratory values to vital signs did not significantly improve its overall performance.

Early prediction model of organ/space surgical site infection after elective gastrointestinal or hepatopancreatobiliary cancer surgery.

BACKGROUND: Organ/space SSI is a significant clinical problem. However, early detection of organ/space SSI is difficult, and previous predictive models are limited in their prognostic ability. We aimed to develop and validate a prediction model of organ/space surgical site infection (SSI) using postoperative day 3 laboratory data in patients who underwent gastrointestinal or hepatopancreatobiliary cancer resection. METHODS: This retrospective cohort study using a single-center hospital data from April 2013 to September 2017 included all adult patients who underwent elective gastrointestinal or hepatopancreatobiliary cancer resection. The primary outcome was a presence of organ/space SSI including anastomotic leakage, pancreatic fistula, biliary fistula, or intra-abdominal abscess. We developed and validated a logistic regression model to predict organ/space SSI using laboratory data on postoperative day (POD) 3. Similar models using laboratory data on POD 1 or 5 were developed to compare the predictive ability of each model. RESULTS: A total of 1578 patients were included. Organ/space SSI was diagnosed in 107 patients, with median diagnosis days of 6 (interquartile range, 4-9 days) after surgery. A prediction model using five commonly measured variables on POD 3 was created with the area under the curve (AUC) of 0.883 (95%CI 0.819-0.946). The AUC of a model with POD 1 laboratory data was 0.751 (95%CI 0.655-0.848), while that of POD 5 laboratory data was 0.818 (95%CI 0.730-0.906). CONCLUSIONS: Laboratory data on POD 3 could forecast organ/space SSI precisely. Further prospective studies are warranted to investigate the clinical impact of this model.

Rapidly growing neuroendocrine carcinoma of the gallbladder: A case report.

Gallbladder neuroendocrine carcinomas are rare tumors with a prognosis poorer than that of other gallbladder carcinomas. These tumors are often detected late and are difficult to treat. We present the case of a 68-year-old woman with small-cell gallbladder neuroendocrine carcinoma. Abdominal sonography and dynamic contrast-enhanced MRI performed at different points in time showed rapid growth. Treatment with surgical resection and adjuvant chemotherapy was instituted. In view of the rapid growth of these tumors, suspicious cases should at least be considered for close follow-up with appropriate imaging studies.

A case of protein-losing gastroenteropathy caused by systemic AA amyloidosis secondary to undifferentiated carcinoma of unknown primary origin.

We report the case of a 61-year-old woman with Kartagener syndrome who presented with a 3-month history of chronic watery diarrhoea and severe hypoalbuminaemia. Histopathological examination of duodenum and large intestine biopsies showed amyloid A (AA) amyloid deposition. Scintigraphy and alpha-1 anti-trypsin clearance evaluations revealed protein-losing gastroenteropathy. Computed tomography with contrast and positron emission tomography showed a pelvic mass with multiple para-aortic lymph node enlargement. We suspected protein-losing gastroenteropathy secondary to AA amyloid produced related to malignant tumours. Following tumour resection, histopathological examination of the lesion revealed undifferentiated carcinoma of unknown origin. Postoperatively, the patient's nutritional condition improved. There has been no recurrence of protein-losing gastroenteropathy 6 months postoperatively. This is the first report of protein-losing gastroenteropathy and AA amyloidosis secondary to undifferentiated carcinoma. Early recognition and intervention could increase the likelihood of amyloidosis remission.

Pembrolizumab-induced pneumonitis with a perilymphatic nodular pattern in a lung cancer patient: A radio-pathologic correlation.

A 66-year-old Japanese man with recurrent adenocarcinoma of the lung p-stage IIIA (pT2bN2M0; version 8) on pembrolizumab was present with gradually worsening dyspnea. Although history and physical examination were unremarkable, high-resolution CT showed the perilymphatic distribution of the pembrolizumab-induced pneumonitis. Consistent with the CT result, biopsy revealed the aggregation of the cytotoxic (CD8+) T-lymphocytes around the lymph tracts. Given the clinical, radiological and pathological findings, pembrolizumab-induced pneumonitis was confirmed. The patient was discharged after terminating the pembrolizumab with ameliorated symptoms. This report, in conjunction with existing literature, illustrates the wide variety of the pembrolizumab-induced pneumonitis and bolsters the current understanding of its pathophysiology.