Results Management

Deep-Learning Natural Language Processing Tool Identifies Imaging Findings to Reduce Risk of Delayed Care

Artificial intelligence (AI) reads through imaging reports to alert physicians about findings that need follow-up.

Each year, thousands of imaging reports are generated at hospitals. While primary results address the issue at hand, a full report can contain other significant yet unexpected findings. But without a process for identifying and communicating these findings, patients could be at risk for delayed or missed treatment.

In 2020, Northwestern Medicine set out to develop a solution to this nationwide healthcare issue, using AI to identify imaging reports with lung- and adrenal-related findings.

Tracking Unreported Findings Through Language

When clinicians order imaging, such as a chest X-ray for suspected pneumonia, a radiologist reviews the images and documents any findings in their final report. In addition to addressing the reason for the imaging, radiologists also document identified findings that may be unrelated. The technical, nuanced language of the report and busy clinician workflow can often result in these additional findings being missed. To address the issue, Northwestern Medicine looked to a type of AI called deep-learning natural language processing (NLP).

The first step was training the algorithm to understand clinical interpretations. This involved manually annotating and classifying thousands of imaging reports and then using them to teach the algorithm to look at word patterns to determine if a radiologist was describing a potential adrenal or lung nodule.

The tool can now sort through thousands of reports and identify imaging findings, both expected and incidental, where follow-up is recommended.

Sharing Results

The tool works quickly — the algorithm presents its findings to the ordering physician within minutes of the imaging report being finalized. Through these alerts, physicians can place relevant follow-up orders that are then tracked to completion. Multidisciplinary teams worked to build the NLP algorithm into the Epic electronic medical record, making it easy to integrate into existing workflows and communicate with patients.

In addition to flagging findings that could otherwise go undetected, the algorithm expedites the process for determining next steps in care and allows for faster treatment, improving outcomes for many patients. In one case, the tool identified an adrenal nodule and alerted the ordering physician, who referred the patient to Northwestern Medicine Endocrinology — all within three days.

As of March 2023, the algorithm has reviewed more than 1.2 million imaging reports, with over 62,000 findings identified and tracked.