Information Delivery on Incidental Findings - Pulmonary Nodules



Deliver accessible information to the patient and care team on incidental findings, such as a pulmonary nodules, and subsequent recommendations


Non-Interpretative, Incidental Finding



Define-AI ID



Patient Facing Subpanel

Panel Chair

Mike Moore

Non-Interpretive Panel Chairs

Alexander J. Towbin, Adam Prater

Panel Reviewers

Patient Facing Subpanel


Creative Commons 4.0


Public Comment

Clinical Implementation

Value Proposition

Incidental findings such as pulmonary nodules are commonly encountered during routine imaging exam acquisition and interpretation. Early identification and treatment of pulmonary nodules, in particular, may result in a significant impact on patient outcomes. Yet patients do not always have immediate and accessible information about these findings. Sharing timely and approachable information can empower patients to take a more active role in their care. Additionally, the importance of disseminating this information appropriately is now widely recognized resulting in legislation that may require providers to share information on incidental findings directly with the patients. Based on these trends, there is a prime opportunity for automation that ensures incidental finding information is delivered in the mode of communication preferred by the patient and sets that patient on track to ensure that proper evidence-based action is taken on the incidental finding. 


A 60-year-old male presents to the emergency room with shortness of breath after a long flight. There, the patient receives a computed tomography (CT) scan for suspected pulmonary embolism. The exam was negative for pulmonary embolism, but the radiologist identified an eight millimeter lung nodule. Prior to the exam, the patient selected (opt in when given the opportunity) to receive the information directly when available. Following the radiologist’s report, the algorithm identified key structures of the report describing the incidental finding and prompted the radiologist to send the patient the information in patient-friendly language by their selected mode. Selected modes could potentially include the patient portal, text message, or a phone call from a provider to the patient.  Alternatively, a patient designated caregiver may also receive the information. Follow ups were scheduled to manage the lung nodule and recorded in the health system database. 

Workflow Description

A radiologist interprets images and renders a report (although augmented intelligent solutions are in development to assist in several areas of interpretation). In the case of a positive finding, a notification may be triggered (automatically prompted to the radiologist by an algorithm based on report text) to transmit associated patient-friendly information on the finding to the patient. In this case, the engine identifies descriptors of the lung nodule from the report. Based on these data, the engine suggests a patient-friendly template with text and graphics representing the information delivered by the patient’s preferred mode of receiving this information as soon as possible. Follow-up recommendations are logged and status is tracked (please refer to separate “close the loop” use case) to ensure the patient received proper follow up.

Considerations for Dataset Development

What procedures might prompt this algorithm?

Chest CT; Chest Radiograph; any additional imaging modality which includes a portion of the lungs (e.g., abdominal CT, and CT/radiographs of the upper extremity or spine)

Are there different types of reports that should be considered when training the algorithm? 

Chest CT; Chest Radiograph.

What do you consider an incidental pulmonary finding?  

As defined within: Guidelines for Management of Incidental Pulmonary Nodules Detected on CT Images: From the Fleischner Society 2017; Radiology: Volume 284: Number 1—July 2017

How does this algorithm launch?

Automatic prompting of radiologist with radiologist confirmation. Integration of a concurrently running natural language processing algorithm (incorporated in radiologist voice dictation) to prompt the radiologist would be of great value to increase radiologist efficiency and confirm likelihood of compliance – an augmented intelligence solution.

Technical Specifications




Radiologist Report


CT, Chest; Radiograph, Chest


Data Type

Structured Text


CT, Radiograph

Body Region


Anatomic Focus


Primary Outputs


Patient-Friendly Template


Present to incidental lung nodule finding information and future steps required in a patient friendly template and distribute in the manner selected by patient and patient designee (patient portal, text based, or provider phone call)

Future Development Ideas

Ideally this algorithm would track patient’s follow up recommendations around incidental findings. This database could be fixed with a messaging system to notify patients and care team to schedule or attend appointments and help close the loop around managing pulmonary nodules to ensure a follow through with recommendations.