Radiologist-AI Double Reading for Incidental Oncology Findings: Detecting Overlooked Metastatic Clues in Routine Chest and Abdomen CT
Main Article Content
Abstract
Background: Routine contrast-enhanced chest and abdomen CT examinations performed for non-oncologic indications may contain subtle incidental oncologic findings that are missed or insufficiently emphasized during focused reporting. Objective: To evaluate whether a radiologist–AI double-reading quality assurance pathway can identify overlooked incidental oncologic findings on routine non-oncology CT and determine their clinical significance. Methods: This retrospective clinical correlation study included 2,486 eligible contrast-enhanced chest, abdomen, or combined chest–abdomen CT examinations from 3,214 retrieved studies at a tertiary-care hospital in Islamabad, Pakistan. De-identified scans were screened using a sensitivity-oriented AI triage system for suspicious skeletal, nodal, and visceral abnormalities. AI-flagged examinations underwent independent radiologist re-review, consensus adjudication, comparison with initial reports, and clinical validation through histopathology, follow-up imaging progression, or strong clinic-oncologic correlation. Results: AI flagged 412 of 2,486 examinations (16.6%). Radiologist re-review confirmed suspicious incidental oncologic findings in 187 flagged cases (45.4%). Of these, 56 cases (29.9%) were discordant with the initial report, representing 2.3% of the total cohort. Skeletal lesions were the most frequent overlooked abnormality (24/56; 42.9%), followed by nodal disease (17/56; 30.4%) and visceral lesions (15/56; 26.8%). Clinical validation confirmed malignancy-related disease in 38 of 56 discordant cases (67.9%). Conclusion: A radiologist–AI double-reading pathway selectively identified clinically meaningful overlooked oncologic findings while limiting secondary review to a minority of routine CT examinations.
Article Details
Section
This work is licensed under a Creative Commons Attribution 4.0 International License.
How to Cite
References
1. Geijer H, Geijer M. Added value of double reading in diagnostic radiology, a systematic review. Insights Imaging. 2018;9:287–301. doi:10.1007/s13244-018-0599-0.
2. Lumbreras B, Donat L, Hernández-Aguado I. Incidental findings in imaging diagnostic tests: a systematic review. Br J Radiol. 2010;83(988):276–289. doi:10.1259/bjr/98067945.
3. O’Sullivan JW, Muntinga T, Grigg S, Ioannidis JPA. Prevalence and outcomes of incidental imaging findings: umbrella review. BMJ. 2018;361:k2387. doi:10.1136/bmj.k2387.
4. MacMahon H, Naidich DP, Goo JM, Lee KS, Leung ANC, Mayo JR, et al. Guidelines for management of incidental pulmonary nodules detected on CT images: Fleischner Society 2017. Radiology. 2017;284(1):228–243. doi:10.1148/radiol.2017161659.
5. Noguchi T, Kido A, Hashimoto N, et al. Deep learning-based algorithm improved radiologists’ performance in bone metastases detection on CT. Eur Radiol. 2022;32:5952–5964. doi:10.1007/s00330-022-08741-3.
6. Zhang Y, Liu X, Chen J, et al. A clinically applicable AI system for detection and diagnosis of bone metastases using CT scans. Nat Commun. 2025;16:9433. doi:10.1038/s41467-025-59433-7.
7. Fu L, Plecha D, Sommer F, et al. System-driven longitudinal follow-up of incidental imaging findings: an initiative with workflow changes. Br J Radiol. 2023;96:20220573. doi:10.1259/bjr.20220573.
8. Fu L, Prevedello LM, Reis M, et al. Implementation and evaluation of dictation macro templates to standardize radiology report recommendations. J Digit Imaging. 2025;38:1–10. doi:10.1007/s10278-024-01357-7.
9. Zarour A, Arpasi P, Song J, et al. The FIND Program: improving follow-up of incidental imaging findings. J Digit Imaging. 2023;36:1268–1276. doi:10.1007/s10278-023-00780-6.
10. Vey B, Gatidis S, Ruppert J, et al. Natural language processing to identify critical findings in radiology reports: a systematic evaluation. J Digit Imaging. 2023;36:360–370. doi:10.1007/s10278-022-00712-w.
11. Polites SF, Habermann EB, Chung AE, et al. A closed-loop system for incidental imaging findings: improving communication and follow-up. J Digit Imaging. 2023;36:1130–1138. doi:10.1007/s10278-022-00773-x.
12. Del Gaizo AJ, Allen BC. Actionable incidental findings on radiologic examinations: existing approaches and future directions. J Radiol Nurs. 2024;43(2):e1–e7. doi:10.1016/j.jradnu.2023.12.003.
13. Evans CS, Arthur R, Kane M, et al. Incidental radiology findings on CT studies in emergency department patients: systematic review and meta-analysis. Ann Emerg Med. 2022;80(3):243–256. doi:10.1016/j.annemergmed.2022.03.027.
14. Kroczek EK, Wieners G, Steffen I, et al. Non-traumatic incidental findings in patients undergoing whole-body CT at initial emergency admission. Emerg Med J. 2017;34(10):643–649. doi:10.1136/emermed-2015-205409.
15. Doughlin GK, Lee SW. Incidental findings in blunt trauma patients: prevalence, follow-up recommendations, and compliance. Emerg Radiol. 2017;24:425–431. doi:10.1007/s10140-017-1479-5.
16. Chapman S, Chadwick C. Expecting the unexpected: incidental findings at a level 1 trauma center. Emerg Radiol. 2023;30:593–602. doi:10.1007/s10140-023-02138-5.
17. Itri JN, Patel SH. Heuristics and cognitive error in medical imaging. AJR Am J Roentgenol. 2018;210(5):1097–1105. doi:10.2214/AJR.17.18907.
18. Bruno MA, Walker EA, Abujudeh HH. Error and discrepancy in radiology: inevitable or avoidable? Insights Imaging. 2015;6:171–182. doi:10.1007/s13244-016-0534-1.
19. Pesapane F, Gnocchi G, Quarrella C, et al. Errors in radiology: a standard review. J Clin Med. 2024;13(15):4306. doi:10.3390/jcm13154306.
20. Van Diepen S, van Lieshout EMM, Wijffels MME, van Walsum T. A deep learning-based approach to automated rib fracture detection and classification on CT. Int J Comput Assist Radiol Surg. 2025;20:1–12. doi:10.1007/s11548-025-03390-5.