Analysis of radiotherapy planning images using colour blindness filters

Colour vision deficiency (CVD), or colour blindness, affects the interpretation of colours and poses a significant risk in clinical settings that rely on colour-coded information. This study aims to quantitatively assess the impact of CVD on the interpretation of radiotherapy treatment planning images, in which isodose curves and target volumes are represented by a specific colour spectrum.

By applying simulated filters for deuteranopia, protanopia, and tritanopia to a set of radiotherapy images, a quantitative analysis framework was developed using Python. This framework evaluates not only global changes but also functional and perceptual integrity through advanced metrics: Intersection over Union (IoU) for spatial accuracy of the planning target volume (PTV), structural similarity index (SSIM) for overall image structure, and CIEDE2000 (Delta E) for perceptual colour difference.

Our results demonstrate that deuteranopia and protanopia consistently lead to a complete failure in identifying the red-coded PTV (IoU = 0.0) across all tested images, despite the overall image structure remaining deceptively intact (SSIM > 0.98). The Delta E metric confirmed a massive perceptual shift in color for these conditions, with values ranging from 22.15 to 29.93. In contrast, tritanopia showed significantly better PTV identification (IoU ranging from 0.7256 to 0.9493) and lower perceptual colour differences (Delta E ranging from 4.47 to 11.64).

These findings quantify a severe clinical risk, highlighting the urgent need for color-accessible radiotherapy planning software to ensure patient safety.