Speed, Accuracy, Confidence - Diagnostic Process in Radiology Interpretation
- Observation, Reflection and One Future Proposal (for development)
Poh-Sun Goh
Friday 16 April, 2021; 1420pm and 1615pm, Singapore Time (first shared by me with consultant at NTFGH, and then group of radiology residents at NTFGH), and Sunday 18 April 2021 @0600am, Singapore Time)
Our professional role in Radiology is to review images, declare that no significant abnormality is 'seen', or that the imaging study or scan 'appears' normal, or detect, categorise-diagnose-label-name an abnormality (or abnormalities), and justify this opinion-assessment (through a description of key features, including matching this opinion with given information (usually from that provided on the imaging request form).
With progressive experience, training, and feedback, Radiologists develop increasing speed, accuracy and confidence in making and giving a radiological diagnosis-opinion, and justifying this.
With digital imaging and use of digital platforms to support the image review, and radiology report transcription and delivery process, integrated into the whole electronic medical record, the possibility of dynamic peer observation, review and assessment of this process, and both peer and increasingly in the future AI (artificial or augmented intelligence) assistance is likely.
Imagine how AI (and AI assisted human coaching and training) might add value to this process by reflecting on observation of how trainees and trained experienced Radiology consultants perform in day to day practice. Notice the practitioner or trainee 'pauses' when reviewing or scrolling through digital images or image sets, and interacts with these images when 'something' 'catches' their eye. This gives an observer, or an AI system, insight into the 'cognitive process', of what is going on in the practitioner's or trainee's mind.
One can ('We' can) imagine a future where the 'digital platform' observes, makes note of these 'pauses', and 'interactions', and then correlates this with the 'typed radiology opinion or report', and cross references this, dynamically with an ever enlarging 'database' of 'final' clinical discharge summaries, 'proven' final clinical diagnoses, and results of laboratory and pathology tests. Imagine the value of this for both a Radiologist or Trainee; and instructor-teacher-coach.
The development of a functional, validated system to do this would have a transformative effect on Radiology practice, and training. The current generation of trainees, and consultants can explore this proposal further, and drive innovation, and transformation in our practice.
