A novel homogeneity index based on the area integration of dose volume histogram

The Homogeneity Index (HI) is a critical clinical metric for evaluating the conformity of radiation dose distribution to the prescribed dose in the target volume during radiotherapy, where cold spots may reduce tumor control probability and hot spots increase toxicity risks to adjacent critical organs. Conventional HIs based on limited dose-volume histogram (DVH) points fail to distinguish between cases with similar dose values but different spatial distributions, while voxel-based HIs offer higher accuracy but face challenges in clinical adoption due to computational complexity. This study introduces a novel HI through DVH area integration, systematically quantifying both cold and hot spot effects in target volumes to provide an intuitive and clinically practical tool for dose uniformity assessment, with rigorous validation of its dose evaluation accuracy.

The novel HIs comprise three parameters: a cold spot index (HI_c), a hot spot index (HI_h), and a global homogeneity index (HI_g), as defined in Eqs. (1)-(4). HI_c and HI_h quantify the impacts of cold and hot spots on dose homogeneity, with ideal values approaching zero (smaller values indicate better uniformity). This study involved 11 patients with brain metastases who underwent CyberKnife (CK) stereotactic radiosurgery (SRS) with varying prescription isodose lines (PIDLs). The novel HIs were calculated from DVHs to differentiate dose uniformity among SRS plans. Pearson correlations between the novel HIs and corresponding point doses or previously established HIs were also examined.

The novel HIs effectively differentiated dose homogeneity across various PIDLs, with values decreasing as PIDL increased. The maximum variations in the mean HI_c, HI_h, and HI_g were -66.86%, -84.04%, and -83.95%, respectively, for PIDLs ranging from 50% to 90%. Strong negative correlations were observed between the novel HIs (HI_c, HI_h, HI_g) and PIDLs, with Pearson correlation coefficients consistently exceeding -0.9 (P<0.05). Significant inter-PIDLs variations in HI values were identified (P<0.05), most notably for HI_c between 90% and 50%-75% PIDLs groups. Consistent with these findings, both the HI_h and HI_g demonstrated statistically significant variations across different PIDLs (P<0.05). Compared to point-dose-based methods, the novel HIs exhibited significantly stronger correlations with voxel-based dose calculations(all P<0.05). Furthermore, robust associations (P<0.05) were observed between the novel HIs and: (1) corresponding point dose measurements, and (2) conventional HIs from prior studies.

The novel HIs, based on DVH area integration, effectively evaluate dose uniformity while accounting for both cold and hot spots. Significant agreement was observed between the novel HIs and previously established HIs, particularly those based on voxel-dose calculations. The novel HIs are sufficiently sensitive to differentiate the uniformity among various treatment plans and provide a more accurate and reliable tool for dose homogeneity assessment.