Purpose To characterize tumour microstructure through conventional apparent diffusion coefficient (ADC) derived from diffusion-weighted MRI (DWI), in skull-base chordoma (SBC) patients in proton and carbon-ion radiotherapy. Materials and Methods Sixty-nine patients affected by SBC, who underwent conventional DWI before treatment and were enrolled for proton (n=37) or carbon-ions (n=32) therapy at the National Center for Oncological Hadrontherapy (CNAO), were retrospectively selected. Tumour volume contours (GTV), clinically verified local control information (LC) and histological information (Ki-67, proliferation index) was collected. From mono-exponential fits of DWI, ADC maps were estimated using different sets of b-values. DWI signals were simulated from synthetic substrates as reported in a previous study (doi:10.1002/mp.14689), which mimic the cellular packing of a tumour tissue with density and geometrical properties defined by parameters of cell radius (R), volume fraction (vf), and diffusion (D). An optimised version of the original computational framework was proposed and then implemented to produce quantitative maps of R, vf and D from ADCs. Histogram-based metrics (mean, median, interquartile range, entropy) were considered for voxel-wise analyses. Logistic regression models were encapsulated in a leave-pair-out cross-validation routine to define the best patients’ separation in terms of LC and Ki-67. Results After that performance was improved by optimising the method proposed in the literature, a quantitative description of the tumour microstructure was obtained from the measured ADC through the R, vf, D. This brought the successful classification of patients according to treatment outcome and non-invasive verification of the cell proliferation index with superior performance compared to conventional ADC. Conclusions The estimation of microstructural parameters can positively contribute to the prediction of outcome and proliferative capacity in SBC. It can also provide the basis for describing tumour characteristics and effects related to particle therapy on a microscopic scale, implying advantages for patient stratification and treatment optimization.
Obiettivo Caratterizzare la microstruttura tumorale attraverso il coefficiente di diffusione apparente (ADC) in pazienti con cordoma della base cranica (SBC) trattati con adroterapia. Materiali e metodi Sono stati selezionati retrospettivamente sessantanove pazienti affetti da SBC, sottoposti a una DW-MRI convenzionale prima del trattamento radioterapico con protoni (n=37) o con ioni carbonio (n=32) presso il Centro Nazionale di Adroterapia Oncologica (CNAO). Sono stati raccolti i contorni dei volumi tumorali (GTV), informazioni sul controllo locale (LC) e informazioni istologiche (Ki-67, indice di proliferazione). Sono state calcolate le mappe ADC con un fit mono-esponenziale del segnale DWI utilizzando diversi set di b-values. Segnali DWI sono stati simulati da substrati sintetici (realizzati per uno studio precedente, doi:10.1002/mp.14689) che imitano l'impaccamento cellulare di un tessuto tumorale con densità e proprietà geometriche definite da parametri di raggio cellulare (R), frazione volumetrica (vf) e diffusione (D). È stata proposta e poi implementata una versione ottimizzata del framework originale per produrre, dagli ADC, mappe quantitative di R, vf, D. Metriche di statistica descrittiva sono state considerate per le analisi voxel-wise. Le stime sono state date in ingresso a un modello di Regressione Logistica (con Leave Pair Out Cross Validation) per distinguere i pazienti in termini di LC e Ki-67. Risultati Dopo aver ottenuto prestazioni migliori ottimizzando il metodo originale, a partire dall’ADC è stata ottenuta una descrizione quantitativa della microstruttura tumorale in termini di R, vf, D. Ciò ha permesso di classificare con successo i pazienti in base all'esito del trattamento e di verificare in modo non invasivo l'indice di proliferazione con performance superiori al convenzionale ADC. Conclusioni I parametri microstrutturali stimati possono contribuire alla predizione dell’outcome e delle capacità proliferative in SBC, e forniscono uno strumento per la descrizione del volume tumorale e degli effetti dei trattamenti con particelle su scala microscopica, implicando vantaggi per la stratificazione dei pazienti e l'ottimizzazione del trattamento.
Modello microstrutturale a partire da DW-MRI in pazienti affetti da cordoma della base cranica trattati con radioterapia a particelle
Morelli, Letizia
2019/2020
Abstract
Purpose To characterize tumour microstructure through conventional apparent diffusion coefficient (ADC) derived from diffusion-weighted MRI (DWI), in skull-base chordoma (SBC) patients in proton and carbon-ion radiotherapy. Materials and Methods Sixty-nine patients affected by SBC, who underwent conventional DWI before treatment and were enrolled for proton (n=37) or carbon-ions (n=32) therapy at the National Center for Oncological Hadrontherapy (CNAO), were retrospectively selected. Tumour volume contours (GTV), clinically verified local control information (LC) and histological information (Ki-67, proliferation index) was collected. From mono-exponential fits of DWI, ADC maps were estimated using different sets of b-values. DWI signals were simulated from synthetic substrates as reported in a previous study (doi:10.1002/mp.14689), which mimic the cellular packing of a tumour tissue with density and geometrical properties defined by parameters of cell radius (R), volume fraction (vf), and diffusion (D). An optimised version of the original computational framework was proposed and then implemented to produce quantitative maps of R, vf and D from ADCs. Histogram-based metrics (mean, median, interquartile range, entropy) were considered for voxel-wise analyses. Logistic regression models were encapsulated in a leave-pair-out cross-validation routine to define the best patients’ separation in terms of LC and Ki-67. Results After that performance was improved by optimising the method proposed in the literature, a quantitative description of the tumour microstructure was obtained from the measured ADC through the R, vf, D. This brought the successful classification of patients according to treatment outcome and non-invasive verification of the cell proliferation index with superior performance compared to conventional ADC. Conclusions The estimation of microstructural parameters can positively contribute to the prediction of outcome and proliferative capacity in SBC. It can also provide the basis for describing tumour characteristics and effects related to particle therapy on a microscopic scale, implying advantages for patient stratification and treatment optimization.| File | Dimensione | Formato | |
|---|---|---|---|
|
2021_04_Morelli.pdf
non accessibile
Descrizione: Testo tesi
Dimensione
11.77 MB
Formato
Adobe PDF
|
11.77 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/174118