An effective calculation method for an overlap volume histogram descriptor and its application in IMRT plan retrieval

Area of Science:

• Medical Physics
• Radiation Oncology
• Computational Imaging

Background:

• Intensity modulated radiation treatment (IMRT) planning requires accurate representation of complex geometrical relationships between tumors and organs at risk (OARs).
• Retrieving similar past IMRT plans can optimize current treatment planning, but requires robust similarity metrics.
• Overlap Volume Histogram (OVH) descriptors offer a potential method for quantifying these geometrical relationships.

Purpose of the Study:

• To develop and validate a morphology-based method for calculating OVH descriptors.
• To introduce a novel similarity measurement for retrieving similar IMRT plans based on OVH descriptors.
• To assess the effectiveness of the proposed method in improving IMRT plan retrieval for nasopharyngeal carcinoma (NPC) cases.

Main Methods:

• A morphology-based approach using dilation or erosion operators was employed for OVH descriptor calculation.
• Minimum and maximum distances between tumors and OARs determined contraction/expansion parameters for OVH computation.
• Similarity was measured by the area between OVH descriptors, with 3D reconstructions for visual validation.

Main Results:

• The proposed method effectively calculated OVH descriptors that accurately reflect 3D geometrical features of tumors and OARs.
• IMRT plan retrieval based on OVH similarity showed good agreement with visual inspection of 3D reconstructions.
• An inverse correlation was observed between the area of OVH descriptors and visual similarity, indicating method efficacy.

Conclusions:

• The developed morphology-based method provides an effective means for calculating OVH descriptors.
• The novel similarity measurement based on OVH descriptors facilitates accurate retrieval of similar IMRT cases.
• This approach holds promise for optimizing IMRT planning by leveraging historical treatment data.