In medical imaging and magnetic resonance imaging (MRI), images generally represent the interaction between electromagnetic waves and the human body, often provided in multiple modalities, each represented in gray scale. However, the analysis and interpretation of these images mainly occur sequentially or, as in the case of automated tools, as a con-catenation of independent sources of information. The nonlinear relationships between them are not exploited or left to the learning process of a complex automated strategy. In contrast, combining multiple modalities into pseudo-color images could enable the exploitation of nonlinear relationships due to color perception and color contrast, which play a crucial role in human vision to recognize objects effectively and efficiently. In principle, this can be extended to automated systems. However, automatic strategies do not fully exploit color representation if these nonlinear relationships are not modeled. In this paper, we show how a graph model of a device-independent color encoding (CIE XYZ) supplemented with a metric in the form of Euclidean distance between colors can become an effective tool for image segmentation from multiple sources. Tests were conducted on multimodal brain MRIs collected in a public database. The results demonstrate the importance of a graph model for mimicking color relationships, which could be very useful in medical image analysis and interpretation and other color-based computer vision applications.

Graph Model to Represent Color Closeness in Pseudo-color Multimodal MRI

Polsinelli M.;
2023-01-01

Abstract

In medical imaging and magnetic resonance imaging (MRI), images generally represent the interaction between electromagnetic waves and the human body, often provided in multiple modalities, each represented in gray scale. However, the analysis and interpretation of these images mainly occur sequentially or, as in the case of automated tools, as a con-catenation of independent sources of information. The nonlinear relationships between them are not exploited or left to the learning process of a complex automated strategy. In contrast, combining multiple modalities into pseudo-color images could enable the exploitation of nonlinear relationships due to color perception and color contrast, which play a crucial role in human vision to recognize objects effectively and efficiently. In principle, this can be extended to automated systems. However, automatic strategies do not fully exploit color representation if these nonlinear relationships are not modeled. In this paper, we show how a graph model of a device-independent color encoding (CIE XYZ) supplemented with a metric in the form of Euclidean distance between colors can become an effective tool for image segmentation from multiple sources. Tests were conducted on multimodal brain MRIs collected in a public database. The results demonstrate the importance of a graph model for mimicking color relationships, which could be very useful in medical image analysis and interpretation and other color-based computer vision applications.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11386/4896675
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 1
social impact