Farrah Jabeen MBBCh, Presenter: Nothing to Disclose

Robert W. Dalley MD, Abstract Co-Author: Nothing to Disclose

Yoshimi Anzai MD, Abstract Co-Author: Nothing to Disclose

Quality reports should include not only description of radiological findings but also diagnosis and recommendation based on the diagnosis. BI-RAD used for screening mammography is one of the examples to clarify the level of suspicion for presence of malignancy directly related to recommendation. Purpose of this study is to apply neuro-Imaging reporting and data system (NI-RAD) for head and neck cancer model to test feasibility and to assess inter-observer reliability.

The study was approved by the institutional review board. 40 consecutive neck CTs performed for patients presenting with neck mass were selected for this preliminary analysis. Patients with known history of head and neck cancer were excluded. Two expert head and neck radiologists reviewed the images independently without knowledge of final pathology and coded the study as following: 1) Normal, 2) Definitely benign, 3) Probably benign, 4) Reasonable probability of being malignant, and 5) Highly suspicious for malignancy. Final pathology, clinical examinations, or follow-up were used to confirm the final diagnosis.

Sensitivity, specificity, and accuracy of reader A were 100%, 65.7%, and 70%, respectively. Those of reader B were 100%, 60%, and 65%, respectively. The area under the ROC (Receiver Operating Characteristics) curves for reader A and B were, 0.971 (95% CI 0.93-1.00) and 0.877 (95%CI 0.76 - 0.99). Percent agreement and weighted Kappa between two readers were 83% and 0.566.

The concept of BIRAD is applicable for neck CT for patients presenting with neck mass. The NIRAD clarifies the degree of suspicion for presence of malignancy and improves communication between radiologists and referring physicians. Application of NIRAD also allows quantification of diagnostic accuracy.

Application of BIRAD concept is feasible for other areas of oncologic imaging. NIRAD helps assess diagnostic accuracy of each reader and is potentially used for QA/QI programs.

Jabeen, F, Dalley, R, Anzai, Y, NeuroImaging Reporting and Data System (NI-RAD): Initial Application for Head and Neck Cancer.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8012965.html