Temporal bone radiology report classification using open source machine learning and natural langue processing libraries

Artificial Intelligence in Temporal Bone Imaging: A Systematic Review

OBJECTIVE

The human temporal bone comprises more than 30 identifiable anatomical components. With the demand for precise image interpretation in this complex region, the utilization of artificial intelligence (AI) applications is steadily increasing. This systematic review aims to highlight the current role of AI in temporal bone imaging.

DATA SOURCES

A Systematic Review of English Publications searching MEDLINE (PubMed), COCHRANE Library, and EMBASE.

REVIEW METHODS

The search algorithm employed consisted of key items such as 'artificial intelligence,' 'machine learning,' 'deep learning,' 'neural network,' 'temporal bone,' and 'vestibular schwannoma.' Additionally, manual retrieval was conducted to capture any studies potentially missed in our initial search. All abstracts and full texts were screened based on our inclusion and exclusion criteria.

RESULTS

A total of 72 studies were included. 95.8% were retrospective and 88.9% were based on internal databases. Approximately two-thirds involved an AI-to-human comparison. Computed tomography (CT) was the imaging modality in 54.2% of the studies, with vestibular schwannoma (VS) being the most frequent study item (37.5%). Fifty-eight out of 72 articles employed neural networks, with 72.2% using various types of convolutional neural network models. Quality assessment of the included publications yielded a mean score of 13.6 ± 2.5 on a 20-point scale based on the CONSORT-AI extension.

CONCLUSION

Current research data highlight AI's potential in enhancing diagnostic accuracy with faster results and decreased performance errors compared to those of clinicians, thus improving patient care. However, the shortcomings of the existing research, often marked by heterogeneity and variable quality, underscore the need for more standardized methodological approaches to ensure the consistency and reliability of future data.

LEVEL OF EVIDENCE

NA Laryngoscope, 2024.

Automatic text classification of actionable radiology reports of tinnitus patients using bidirectional encoder representations from transformer (BERT) and in-domain pre-training (IDPT)

Background

Given the increasing number of people suffering from tinnitus, the accurate categorization of patients with actionable reports is attractive in assisting clinical decision making. However, this process requires experienced physicians and significant human labor. Natural language processing (NLP) has shown great potential in big data analytics of medical texts; yet, its application to domain-specific analysis of radiology reports is limited.

Objective

The aim of this study is to propose a novel approach in classifying actionable radiology reports of tinnitus patients using bidirectional encoder representations from transformer BERT-based models and evaluate the benefits of in domain pre-training (IDPT) along with a sequence adaptation strategy.

Methods

A total of 5864 temporal bone computed tomography(CT) reports are labeled by two experienced radiologists as follows: (1) normal findings without notable lesions; (2) notable lesions but uncorrelated to tinnitus; and (3) at least one lesion considered as potential cause of tinnitus. We then constructed a framework consisting of deep learning (DL) neural networks and self-supervised BERT models. A tinnitus domain-specific corpus is used to pre-train the BERT model to further improve its embedding weights. In addition, we conducted an experiment to evaluate multiple groups of max sequence length settings in BERT to reduce the excessive quantity of calculations. After a comprehensive comparison of all metrics, we determined the most promising approach through the performance comparison of F1-scores and AUC values.

Results

In the first experiment, the BERT finetune model achieved a more promising result (AUC-0.868, F1-0.760) compared with that of the Word2Vec-based models(AUC-0.767, F1-0.733) on validation data. In the second experiment, the BERT in-domain pre-training model (AUC-0.948, F1-0.841) performed significantly better than the BERT based model(AUC-0.868, F1-0.760). Additionally, in the variants of BERT fine-tuning models, Mengzi achieved the highest AUC of 0.878 (F1-0.764). Finally, we found that the BERT max-sequence-length of 128 tokens achieved an AUC of 0.866 (F1-0.736), which is almost equal to the BERT max-sequence-length of 512 tokens (AUC-0.868,F1-0.760).

Conclusion

In conclusion, we developed a reliable BERT-based framework for tinnitus diagnosis from Chinese radiology reports, along with a sequence adaptation strategy to reduce computational resources while maintaining accuracy. The findings could provide a reference for NLP development in Chinese radiology reports.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-022-01946-y.

Closing the loop: automatically identifying abnormal imaging results in scanned documents

OBJECTIVES

Scanned documents (SDs), while common in electronic health records and potentially rich in clinically relevant information, rarely fit well with clinician workflow. Here, we identify scanned imaging reports requiring follow-up with high recall and practically useful precision.

MATERIALS AND METHODS

We focused on identifying imaging findings for 3 common causes of malpractice claims: (1) potentially malignant breast (mammography) and (2) lung (chest computed tomography [CT]) lesions and (3) long-bone fracture (X-ray) reports. We train our ClinicalBERT-based pipeline on existing typed/dictated reports classified manually or using ICD-10 codes, evaluate using a test set of manually classified SDs, and compare against string-matching (baseline approach).

RESULTS

A total of 393 mammograms, 305 chest CT, and 683 bone X-ray reports were manually reviewed. The string-matching approach had an F1 of 0.667. For mammograms, chest CTs, and bone X-rays, respectively: models trained on manually classified training data and optimized for F1 reached an F1 of 0.900, 0.905, and 0.817, while separate models optimized for recall achieved a recall of 1.000 with precisions of 0.727, 0.518, and 0.275. Models trained on ICD-10-labelled data and optimized for F1 achieved F1 scores of 0.647, 0.830, and 0.643, while those optimized for recall achieved a recall of 1.0 with precisions of 0.407, 0.683, and 0.358.

DISCUSSION

Our pipeline can identify abnormal reports with potentially useful performance and so decrease the manual effort required to screen for abnormal findings that require follow-up.

CONCLUSION

It is possible to automatically identify clinically significant abnormalities in SDs with high recall and practically useful precision in a generalizable and minimally laborious way.

A systematic review of natural language processing applied to radiology reports

BACKGROUND

Natural language processing (NLP) has a significant role in advancing healthcare and has been found to be key in extracting structured information from radiology reports. Understanding recent developments in NLP application to radiology is of significance but recent reviews on this are limited. This study systematically assesses and quantifies recent literature in NLP applied to radiology reports.

METHODS

We conduct an automated literature search yielding 4836 results using automated filtering, metadata enriching steps and citation search combined with manual review. Our analysis is based on 21 variables including radiology characteristics, NLP methodology, performance, study, and clinical application characteristics.

RESULTS

We present a comprehensive analysis of the 164 publications retrieved with publications in 2019 almost triple those in 2015. Each publication is categorised into one of 6 clinical application categories. Deep learning use increases in the period but conventional machine learning approaches are still prevalent. Deep learning remains challenged when data is scarce and there is little evidence of adoption into clinical practice. Despite 17% of studies reporting greater than 0.85 F1 scores, it is hard to comparatively evaluate these approaches given that most of them use different datasets. Only 14 studies made their data and 15 their code available with 10 externally validating results.

CONCLUSIONS

Automated understanding of clinical narratives of the radiology reports has the potential to enhance the healthcare process and we show that research in this field continues to grow. Reproducibility and explainability of models are important if the domain is to move applications into clinical use. More could be done to share code enabling validation of methods on different institutional data and to reduce heterogeneity in reporting of study properties allowing inter-study comparisons. Our results have significance for researchers in the field providing a systematic synthesis of existing work to build on, identify gaps, opportunities for collaboration and avoid duplication.

Comprehensive Word-Level Classification of Screening Mammography Reports Using a Neural Network Sequence Labeling Approach

Radiology reports contain a large amount of potentially valuable unstructured data. Recently, neural networks have been employed to perform classification of radiology reports over a few classes at the document level. The success of neural networks in sequence-labeling problems such as named entity recognition and part of speech tagging suggests that they could be used to classify radiology report text with greater granularity. We employed a neural network architecture to comprehensively classify mammography report text at the word level using a sequence labeling approach. Two radiologists devised a comprehensive classification system for screening mammography reports. Each word in each report was manually categorized by a radiologist into one of 33 categories according to the classification system. Tagged words referencing the same finding were grouped into unique sets. We pre-labeled reports with a rule-based algorithm and then manually edited these annotations for 6705 screening mammography reports (25.1%, 66.8%, and 8.1% BI-RADS 0, 1, and 2, respectively). A combined convolutional and recurrent neural network model was used to label words in each sentence of the individual reports. A siamese recurrent neural network was then used to group findings into sets. Performance of the neural network-based method was compared to a rule-based algorithm and a conditional random field (CRF) model. Global accuracy (percentage of documents where all word tags were predicted correctly) and keyword accuracy (percentage of all words that were labeled correctly, excluding words tagged as unimportant) were calculated on an unseen 519 report test set. Two-tailed t tests were used to assess differences between algorithm performance, and p < 0.05 was used to determine statistical significance. The neural network-based approach showed significantly higher global accuracy compared to both the rule-based algorithm (88.3 vs 57.0%, p < 0.001) and the CRF model (88.3% vs. 75.8%, p < 0.001). The neural network also showed significantly higher keyword level accuracy compared to the rule-based algorithm (95.5% vs. 80.9% p < 0.001) and CRF model (95.5% vs. 76.9%, p < 0.001). We demonstrate the potential of neural networks to accurately perform word-level multilabel classification of free text radiology reports across 33 classes, thus showing the utility of a sequence labeling approach to NLP of radiology reports. We found that a neural network classifier outperforms a rule-based algorithm and a CRF classifier for comprehensive multilabel classification of free text screening mammography reports at the word level. By approaching radiology report classification as a sequence-labeling problem, we demonstrate the ability of neural networks to extract data from free text radiology reports at a level of granularity not previously reported.

Deep learning to convert unstructured CT pulmonary angiography reports into structured reports

Background

Structured reports have been shown to improve communication between radiologists and providers. However, some radiologists are concerned about resultant decreased workflow efficiency. We tested a machine learning-based algorithm designed to convert unstructured computed tomography pulmonary angiography (CTPA) reports into structured reports.

Methods

A self-supervised convolutional neural network-based algorithm was trained on a dataset of 475 manually structured CTPA reports. Labels for individual statements included “pulmonary arteries,” “lungs and airways,” “pleura,” “mediastinum and lymph nodes,” “cardiovascular,” “soft tissues and bones,” “upper abdomen,” and “lines/tubes.” The algorithm was applied to a test set of 400 unstructured CTPA reports, generating a predicted label for each statement, which was evaluated by two independent observers. Per-statement accuracy was calculated based on strict criteria (algorithm label counted as correct if the statement unequivocally contained content only related to that particular label) and a modified criteria, accounting for problematic statements, including typographical errors, statements that did not fit well into the classification scheme, statements containing content for multiple labels, etc.

Results

Of the 4,157 statements, 3,806 (91.6%) and 3,986 (95.9%) were correctly labeled by the algorithm using strict and modified criteria, respectively, while 274 (6.6%) were problematic for the manual observers to label, the majority of which (n = 173) were due to more than one section being included in one statement.

Conclusion

This algorithm showed high accuracy in converting free-text findings into structured reports, which could improve communication between radiologists and clinicians without loss of productivity and provide more structured data for research/data mining applications.

Autonomous detection, grading, and reporting of postoperative complications using natural language processing

INTRODUCTION

Natural language processing, a computer science technique that allows interpretation of narrative text, is infrequently used to identify surgical complications. We designed a natural language processing algorithm to identify and grade the severity of deep venous thrombosis and pulmonary embolism (together: venous thromboembolism).

METHODS

Patients from our 2011-2014 American College of Surgeons National Surgical Quality Improvement Project cohorts with a duplex ultrasound or a computerized tomography angiography of the chest performed within 30 days of surgery were divided into training and validation datasets. A "bag of words" approach classified the reports; other electronic health record data classified the venous thromboembolism's severity.

RESULTS

Of the 10,295 American College of Surgeons National Surgical Quality Improvement Project patients, 251 were used in our deep venous thromboses validation cohort (273 total ultrasounds) and 506 in our pulmonary embolisms cohort (552 total computerized tomography angiographies). For deep venous thromboses the sensitivity and specificity were 85.1% and 94.6%, while for pulmonary embolisms they were 90% and 98.7%. Most discordances were due to lack of imaging documentation of a deep venous thrombosis (28/41, 68.3%) or pulmonary embolism (6/6, 100%). Most deep venous thromboses (28 patients, 54.6%) and pulmonary embolisms (25 patients, 75.8%) required administration of therapeutic intravenous or subcutaneous anticoagulation.

CONCLUSION

Natural language processing can reliably detect the presence of postoperative venous thromboembolisms, and its use should be expanded for the detection of other conditions from narrative documentation.

Making Sense of Big Textual Data for Health Care: Findings from the Section on Clinical Natural Language Processing

Objectives: To summarize recent research and present a selection of the best papers published in 2016 in the field of clinical Natural Language Processing (NLP). Method: A survey of the literature was performed by the two section editors of the IMIA Yearbook NLP section. Bibliographic databases were searched for papers with a focus on NLP efforts applied to clinical texts or aimed at a clinical outcome. Papers were automatically ranked and then manually reviewed based on titles and abstracts. A shortlist of candidate best papers was first selected by the section editors before being peer-reviewed by independent external reviewers. Results: The five clinical NLP best papers provide a contribution that ranges from emerging original foundational methods to transitioning solid established research results to a practical clinical setting. They offer a framework for abbreviation disambiguation and coreference resolution, a classification method to identify clinically useful sentences, an analysis of counseling conversations to improve support to patients with mental disorder and grounding of gradable adjectives. Conclusions: Clinical NLP continued to thrive in 2016, with an increasing number of contributions towards applications compared to fundamental methods. Fundamental work addresses increasingly complex problems such as lexical semantics, coreference resolution, and discourse analysis. Research results translate into freely available tools, mainly for English.