Analysis of intratumor heterogeneity robustness derived from FDG-PET in mantle cell lymphoma patients enrolled in the Lyma trial
Creators
- 1. Service de medecine nucleaire, centre hospitalo-universitaire, Hotel-Dieu, CHU de Nantes, 1, place Alexis-Ricordeau, 44093 Nantes cedex 01 (France)
- 2. Inserm UMR 1101, LaTIM, 29609 Brest (France)
- 3. Service d'hematologie, centre hospitalo-universitaire, Hotel-Dieu, CHU de Nantes, 1, place Alexis-Ricordeau, 44093 Nantes (France)
Description
Objective. - The use of texture analysis of PET images has known a significant growth in recent years. The objective of this study was to identify the most robust and relevant textural features indices for patients with mantle cell lymphoma (MCL), in which the use of the SUV remains controversial. Materials and methods. - Patients enrolled in this study were derived from an ancillary study of the multicenter phase III LyMa trial. The most intense focus was extracted for each patient using three delineations methods: Fuzzy Locally Adaptative Bayesian approach, 40% of the SUVmax and a threshold of 2.5 of SUV. Fifteen different textural features have been calculated for each lesion for a total of 45 measurements. Results. - FDG-PET data of 92 patients from 28 different centers were included. Five textural feature parameters were very significantly dependent on the segmentation method. All textural features exhibited high inter-correlation and the most inter-correlated parameters were grouped according to their correlation index allowing us to identify four groups. Correlations between SUV and textural features were low. Eight parameters were highly correlated to volume and three of these became independent above a specific volume. Five parameters were found to be independent of the volume of interest in which they were calculated. Conclusion. - Only a few textural features may bring complimentary and non-redundant quantitative information regarding existing indices. (authors)
Availability note (English)
Available from doi: http://dx.doi.org/10.1016/j.mednuc.2016.03.141Abstract (French)
Objectifs. - L'analyse de texture des images TEP connait un essor particulierement important ces dernieres annees. Nous avons souhaite identifier les indices de texture les plus robustes et pertinents chez des patients atteints de lymphome du manteau (LCM), hemopathie pour laquelle l'utilisation de la SUV reste controversee. Materiel et methodes. - Les patients inclus dans cette etude sont issus d'une etude ancillaire a l'essai multicentrique de phase III LyMa. Pour chaque patient, la lesion metaboliquement la plus active a ete segmentee selon 3 differentes methodes : Fuzzy Locally Adaptive Bayesian, 40 % de la SUVmax et un seuil de SUV a 2,5. Quinze parametres de texture differents lies a l'heterogeneite tumorale ont ete calcules pour chaque lesion, soit 45 mesures au total. Resultats. - Les donnees TEP-FDG de 92 patients provenant de 28 centres differents ont ete incluses. Cinq parametres etaient tres significativement dependants de la methode de segmentation. Les 15 differents indices de texture etaient correles entre eux. Les differents indices de texture les plus correles entre eux ont ete regroupes en fonction de leur index de correlation, permettant ainsi d'identifier 4 groupes. Les correlations entre SUV et index de texture etaient faibles. Huit parametres etaient tres correles au volume et trois de ces index devenaient independants au-dela d'un certain volume-seuil. Cinq parametres etaient independants des volumes d'interet dans lesquels ils avaient ete calcules. Conclusion. - Seuls quelques parametres de texture semblent avoir une valeur ajoutee en apportant une information quantitative non redondante et complementaire des indices deja existants. (auteurs)Additional details
Additional titles
- Original title (French)
- Etude de la robustesse des parametres d'heterogeneite en TEP-FDG des lymphomes du manteau inclus dans l'etude LyMa
Identifiers
Publishing Information
- Journal Title
- Medecine Nucleaire. Imagerie Fonctionnelle et Metabolique
- Journal Volume
- 40
- Journal Issue
- no.4
- Journal Page Range
- p. 297-306
- ISSN
- 0928-1258
- CODEN
- MNIMEX
INIS
- Country of Publication
- France
- Country of Input or Organization
- France
- INIS RN
- 53085827
- Subject category
- S62: RADIOLOGY AND NUCLEAR MEDICINE;
- Descriptors DEI
- CORRELATIONS; FLUORODEOXYGLUCOSE; FUZZY LOGIC; IMAGES; LYMPHOMAS; PATIENTS; POSITRON COMPUTED TOMOGRAPHY
- Descriptors DEC
- ANTIMETABOLITES; COMPUTERIZED TOMOGRAPHY; DIAGNOSTIC TECHNIQUES; DISEASES; DRUGS; EMISSION COMPUTED TOMOGRAPHY; IMMUNE SYSTEM DISEASES; MATHEMATICAL LOGIC; NEOPLASMS; TOMOGRAPHY
Optional Information
- Notes
- 48 refs.