Promise and challenges of heart chamber segmentation from noncontrast CT scans using contrastive unpaired image...

To evaluate the feasibility and challenges of heart chamber segmentation from noncontrast CT scans using contrastive unpaired image translation and deep learning-based segmentation.

We developed ChameleonNet, a framework utilizing the Contrastive Unpaired Translation (CUT) network with decoupled contrastive learning (DCL) loss to synthesize noncontrast CT from contrast CT scans. Using annotations of four heart chambers (left atrium (LA), left ventricle (LV), right atrium (RA), and right ventricle (RV)) from contrast scans, we trained a Hausdorff distance loss-enhanced nnU-Net on synthesized noncontrast images. The translation model was trained with 35,538 contrast-enhanced and 37,197 noncontrast CT slices. The segmentation model was trained with 292 synthesized noncontrast scans. Performance was evaluated using Dice similarity coefficient (DSC) and 95th Hausdorff distance (HD95) on 36 synthesized noncontrast scans, and volume agreement on 36 real noncontrast CT scans was assessed using Pearson correlation, mean absolute percentage error (MAPE), and mean percentage error (MPE).

The segmentation model achieved DSC of 0.94 ( $\pm 0.01$ ), 0.91 ( $\pm 0.04$ ), 0.92 ( $\pm 0.03$ ), 0.93 ( $\pm 0.02$ ), and HD95 of 3.63 ( $\pm 1.49$ ), 5.74 ( $\pm 4.08$ ), 5.18 ( $\pm 1.77$ ), and 5.51 ( $\pm 3.21$ ) mm on synthesized noncontrast images for LA, LV, RA, and RV, respectively. On real noncontrast CT scans, Pearson correlations were 0.93, 0.82, 0.87, and 0.89 (all $p<0.001$ ), with MAPE ranging from 9.22% to 20.79%, and MPE ranging from $-12.52\%$ to 4.67%.

ChameleonNet demonstrated feasibility for heart chamber segmentation from noncontrast CT without manual noncontrast annotations. However, volume errors, particularly for LV and RV, indicate that further refinement and validation are needed before clinical use.