[Paper Review] Investigating the Cognitive Processes Involved in Cancer Cell Image Identification

Jennifer S. Trueblood, William R. Holmes+7 more|arXiv (Cornell University)|Sep 19, 2017

AI in cancer detection1 citations

TL;DR

This study uses signal detection theory and the diffusion decision model to compare cognitive processes in cancer cell image identification between novices and experts. It reveals that despite differences in discriminability, both groups show similar response patterns under time pressure and probabilistic cues, suggesting shared decision-making mechanisms in medical image perception.

ABSTRACT

Training individuals to make accurate decisions from medical images is a critical component of education in diagnostic pathology. We describe a joint experimental and computational modeling approach to examine the similarities and differences in the cognitive processes of novice participants and experienced participants (pathology residents and pathology faculty) in cancer cell image identification. For this study we collected a bank of hundreds of digital images that were identified by cell type and classified by difficulty by a panel of expert hematopathologists. The key manipulations in our study included examining the speed-accuracy tradeoff as well as the impact of prior expectations on decisions. In addition, our study examined individual differences in decision-making by comparing task performance to domain general visual ability (as measured using the Novel Object Memory Test (NOMT) (Richler et al., 2017). Using Signal Detection Theory (SDT) and the Diffusion Decision Model (DDM), we found many similarities between expert and novices in our task. While experts tended to have better discriminability, the two groups responded similarly to time pressure (i.e., reduced caution under speed instructions in the DDM) and to the introduction of a probabilistic cue (i.e., increased response bias in the DDM). These results have important implications for training in this area as well as using novice participants in research on medical image perception and decision-making.

Motivation & Objective

• To understand the cognitive mechanisms underlying cancer cell image identification in both novice and expert diagnosticians.
• To investigate how time pressure and prior expectations affect decision-making in medical image perception.
• To examine individual differences in decision performance using domain-general visual ability measures.
• To evaluate the validity of using novice participants in research on medical image perception and decision-making.
• To compare the cognitive strategies of pathology residents and faculty using computational modeling frameworks.

Proposed method

• Collected a large bank of digitally classified cancer cell images, vetted by expert hematopathologists for cell type and difficulty.
• Employed a speed-accuracy tradeoff paradigm to manipulate decision time and assess response caution.
• Applied Signal Detection Theory (SDT) to analyze discriminability and response bias in image identification tasks.
• Used the Diffusion Decision Model (DDM) to model decision dynamics, including drift rate, threshold, and non-decision time.
• Measured domain-general visual ability using the Novel Object Memory Test (NOMT) to assess individual differences.
• Compared model parameters between novice participants and experienced pathology residents and faculty.

Experimental results

Research questions

• RQ1 How do novices and experts differ in their cognitive processing during cancer cell image identification?
• RQ2 How does time pressure affect response caution and decision accuracy in both novice and expert groups?
• RQ3 How do prior expectations (via probabilistic cues) influence response bias in medical image decision-making?
• RQ4 To what extent do individual differences in visual ability predict performance in medical image perception tasks?
• RQ5 Are the cognitive models derived from expert performance generalizable to novice learners in diagnostic pathology?

Key findings

• Experts demonstrated significantly better discriminability (d') than novices in identifying cancer cells from images.
• Both novices and experts reduced response caution under time pressure, as indicated by similar decreases in decision thresholds in the DDM.
• The introduction of probabilistic cues led to increased response bias in both groups, suggesting similar sensitivity to prior expectations.
• Despite differences in accuracy and discriminability, the cognitive dynamics of decision-making were remarkably similar between experts and novices.
• Individual differences in visual ability, as measured by the NOMT, were correlated with performance, though not the primary driver of expert-novice differences.
• The findings suggest that novice participants can be validly used in research on medical image perception, as their decision-making processes align closely with experts under controlled conditions.