Figure 4: Gleason-score adjusted performance of Anisotropy and CAPRA-S. | Scientific Reports

Figure 4: Gleason-score adjusted performance of Anisotropy and CAPRA-S.

From: Prediction of prostate cancer recurrence using quantitative phase imaging: Validation on a general population

Figure 4

(A) At Gleason scores of 5–6, both anisotropy and CAPRA-S show poor performance at identification of recurrent individuals. However, the probability of recurrence at this stage is 3% and CAPRA-S outperforms anisotropy due to it’s ability to identify 100% of the non-recurrent cases. (B) At Gleason score of 7 (3 + 4), anisotropy (AUC 0.72) and CAPRA-S (AUC 0.68) show comparable performance. The probability of recurrence at this stage was 16%. (C) At Gleason score of 7 (4 + 3), the probability of recurrence is 43%. CAPRA-S (AUC 0.38) failed due to overestimation of the recurrence risk. Anisotropy was able to identify recurrence (AUC 0.73). (D) At Gleason score of 8–10, the probability of recurrence is 64%. Anisotropy (AUC 0.73) was able to identify the recurrent cases with a higher level of accuracy than CAPRA-S (AUC 0.6).

Back to article page