Technical Note: New similarity index for radiotherapy and medical imaging
Autor: | Gustavo Figueiredo, Vadim Y. Kuperman |
---|---|
Rok vydání: | 2020 |
Předmět: |
Normalization (statistics)
Index (economics) Current (mathematics) Computer science business.industry Radiotherapy Planning Computer-Assisted Structure (category theory) Radiotherapy Dosage Pattern recognition Technical note General Medicine Similarity measure 030218 nuclear medicine & medical imaging Radiography 03 medical and health sciences 0302 clinical medicine Similarity (network science) 030220 oncology & carcinogenesis Medical imaging Artificial intelligence Element (category theory) Tomography X-Ray Computed business |
Zdroj: | Medical Physics. 47:4325-4331 |
ISSN: | 2473-4209 0094-2405 |
DOI: | 10.1002/mp.14234 |
Popis: | PURPOSE To describe a new similarity index and consider its biomedical applications. METHODS Similarity index for a pair of objects is defined by the number of shared features and total number of features in these objects. Similarity measure for more than two objects is commonly defined by using pairwise similarity indices. In the current study we suggest a novel similarity index which depends on the number of features shared between multiple objects and does not have the limitations of the recently described similarity measures. In order to introduce the new index, we consider a concept of "commonality." For a collection of sets A1,A2…,AN , commonality of a given element equals the number of sets this element belongs to. The similarity index for the compared sets is then defined by a weighted sum of normalized commonalities. RESULTS The considered biomedical applications of the proposed index include comparison of independent delineations of critical cranial structures in MR images and comparison of isodose distributions from different radiotherapy plans. CONCLUSIONS This study describes a novel similarity index which can be used to assess the similarity of multiple independent delineations of the anatomical structure or similarity of multiple dose distributions. Unlike the commonly used pairwise similarity indices, the new index is defined by the number of elements shared between multiple sets. Potential applications of the suggested similarity index for radiotherapy and medical imaging have been described. |
Databáze: | OpenAIRE |
Externí odkaz: |