With the number of cancer survivors increasing, it is important to study treatment effects on quality of life. Chemotherapy is a common treatment strategy, but cognitive impairment has been reported in some patients after treatment, sometimes persisting for months. These symptoms, often called chemobrain, are thought to involve white matter changes, making diffusion MRI a valuable tool to investigate their location and nature. In this thesis, we analyzed diffusion MRI data from three age-matched groups of women: breast cancer patients treated with chemotherapy (CP), patients treated with other therapies (CM), and healthy controls (HC). We applied Constrained Spherical Deconvolution (CSD), which models the diffusion signal as the convolution of a fibre response function and an Orientation Distribution Function (ODF). This enables estimation of fibre orientations even in voxels with complex crossings. Two CSD approaches were used: the state-of-the-art Multi-Tissue CSD (MT-CSD), modeling three tissue types using global response functions, and the newer Local Response Function Estimation in Spherical Deconvolution (LoRE-SD), which estimates response functions locally. LoRE-SD ODFs are unit-normalized and can be scaled to emphasize intra-axonal content, anisotropy, or white matter volume fraction. From these ODFs, we extracted metrics and statistically tested for group differences in white matter microstructure. Results showed that for LoRE-SD findings depended on the ODF scaling. Notably, the fornix exhibited an increase in the metric extracted from the integral of the ODF lobes in CP compared to HC for both MT-CSD and WM-scaled LoRE-SD ODFs, suggesting compensatory mechanisms or network reorganization. However, the fornix is thin and close to the ventricles, making it susceptible to partial voluming and misregistration. This thesis shows that with CSD we can detect white matter changes possibly chemotherapy-related and that LoRE-SD produces results consistent with MT-CSD while offering flexible metrics. Future studies should validate these metrics biologically and investigate longitudinal effects in relation to cognitive assessments.
Con il numero di sopravvissuti al cancro in aumento, è importante studiare gli effetti della terapia sulla qualità della vita. La chemioterapia può provocare deficit cognitivi persistenti, noti come “chemobrain”, legati a modifiche della materia bianca. La risonanza magnetica pesata in diffusione è uno strumento utile per indagare queste alterazioni. In questa tesi sono stati analizzati dati di dMRI di tre gruppi di donne: pazienti con tumore al seno trattate con chemioterapia (CP), pazienti trattate con altre terapie (CM) e controlli sani (HC). Si è applicata la Constrained Spherical Deconvolution (CSD), che modella il segnale di diffusione come la convoluzione tra una funzione di risposta delle fibre e una Orientation Distribution Function (ODF), permettendo di estrarre le direzioni delle fibre anche in voxel con incroci complessi. Sono stati usati due approcci: il Multi-Tissue CSD (MT-CSD), che utilizza funzioni di risposta globali per i principali tessuti, e il più recente Local Response Function Estimation in Spherical Deconvolution (LoRE-SD), che stima funzioni di risposta localmente in ogni voxel. Le ODF di LoRE-SD, normalizzate, possono essere scalate per enfatizzare per esempio il contenuto intra-assonale o la frazione di materia bianca. Da queste ODF sono state estratte metriche, utilizzate in analisi statistiche per rilevare differenze tra i gruppi. Per LoRE-SD, i risultati sono dipesi dal fattore di scala usato per scalare le ODF. La fornice ha mostrato un aumento significativo della metrica derivata dall’integrale dei lobi delle ODF nei CP rispetto a HC sia con MT-CSD sia con LoRE-SD scalando per la frazione di materia bianca, suggerendo possibili meccanismi compensatori o riorganizzazione dei circuiti. Tuttavia, la fornice è sottile e vicina ai ventricoli, quindi suscettibile a effetti di volume parziale e errori di registrazione. Questa tesi dimostra che la CSD può rilevare alterazioni della materia bianca possibilmente legate alla chemioterapia e che LoRE-SD può produrre risultati coerenti con MT-CSD, offrendo metriche flessibili. Studi futuri dovranno validare biologicamente queste metriche e indagare effetti longitudinali con valutazioni cognitive.
White matter degeneration in chemotherapy-related cognitive impairment using constrained spherical deconvolution
Cipolla, Petra
2024/2025
Abstract
With the number of cancer survivors increasing, it is important to study treatment effects on quality of life. Chemotherapy is a common treatment strategy, but cognitive impairment has been reported in some patients after treatment, sometimes persisting for months. These symptoms, often called chemobrain, are thought to involve white matter changes, making diffusion MRI a valuable tool to investigate their location and nature. In this thesis, we analyzed diffusion MRI data from three age-matched groups of women: breast cancer patients treated with chemotherapy (CP), patients treated with other therapies (CM), and healthy controls (HC). We applied Constrained Spherical Deconvolution (CSD), which models the diffusion signal as the convolution of a fibre response function and an Orientation Distribution Function (ODF). This enables estimation of fibre orientations even in voxels with complex crossings. Two CSD approaches were used: the state-of-the-art Multi-Tissue CSD (MT-CSD), modeling three tissue types using global response functions, and the newer Local Response Function Estimation in Spherical Deconvolution (LoRE-SD), which estimates response functions locally. LoRE-SD ODFs are unit-normalized and can be scaled to emphasize intra-axonal content, anisotropy, or white matter volume fraction. From these ODFs, we extracted metrics and statistically tested for group differences in white matter microstructure. Results showed that for LoRE-SD findings depended on the ODF scaling. Notably, the fornix exhibited an increase in the metric extracted from the integral of the ODF lobes in CP compared to HC for both MT-CSD and WM-scaled LoRE-SD ODFs, suggesting compensatory mechanisms or network reorganization. However, the fornix is thin and close to the ventricles, making it susceptible to partial voluming and misregistration. This thesis shows that with CSD we can detect white matter changes possibly chemotherapy-related and that LoRE-SD produces results consistent with MT-CSD while offering flexible metrics. Future studies should validate these metrics biologically and investigate longitudinal effects in relation to cognitive assessments.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_10_Cipolla_Executive_Summary_02.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
4.27 MB
Formato
Adobe PDF
|
4.27 MB | Adobe PDF | Visualizza/Apri |
|
2025_10_Cipolla_Tesi_01.pdf
accessibile in internet per tutti
Descrizione: Testo tesi
Dimensione
19.21 MB
Formato
Adobe PDF
|
19.21 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/10589/243476