Scholarlink

<p>Glioblastoma multiforme (GBM) remains a highly lethal brain malignancy, and although maximal surgical resection can improve outcomes, many tumors are not amenable to aggressive surgery due to proximity to critical neurovascular structures and significant comorbidities. Transcranial focused ultrasound (tcFUS) is a non-invasive alternative which has shown promise in treating neurological diseases, such as treatment of Essential Tremor using thermal ablation. However, for neuro-oncology applications thermal ablations have been constrained by skull heating, long cooling intervals, and limited treatable volumes. Instead, microbubble-enabled FUS approaches, including blood brain barrier opening and drug delivery, have shown promise. Additionally, using microbubbles to non-thermally ablate tissue by inertial microbubble cavitation, causing hemorrhagic lesions, has been investigated pre-clinically. Non-thermal ablation (NTA) uses lower time-averaged powers than thermal ablation, avoiding skull heating issues and enlarging the treatment envelope, as well as lower FUS duty cycle so that larger tumor volumes can potentially be treated. In this work, we quantitatively evaluate the use of magnetic resonance imaging (MRI) for lesion assessment following phase-shift microbubble (PSMB)-enabled tcFUS-mediated NTA in rat model involving seven male Fischer rats, including five healthy rats and two rats bearing F98-LN glioma tumors. We first characterize hemorrhage-related hypointensities in standard T2*-weighted (T2*w) gradient recalled-echo MRI in healthy rat brain across a range of echo times (TE = 5.1–26.0 ms). Ground-truth hemorrhage is derived from registered hematoxylin and eosin (H&amp;E) histology, enabling quantitative comparisons between MRI and histology. We identify a TE = 26 ms as yielding the highest spatial agreement with histology, as measured by soft Dice. To investigate improvements of the immediate outcome prediction, we further introduce a learning-based framework that leverages standard of care multi-parametric MRI (T1w and T2w, in addition to T2*w). Using a simple baseline model as proof of concept, we show that multi-parametric fusion (soft Dice 0.51) can predict histology-derived lesion/hemorrhage patterns 45% higher accuracy than T2*w imaging alone (soft Dice 0.35 using the TE=26 ms echo) in healthy brain. Finally, we demonstrate the robustness of the proposed analysis within the treatment region in a rat brain tumor model, supporting the translational potential of multiparametric MRI-based assessment for non-thermal tcFUS therapies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Histology-validated quantitative MRI assessment of non-thermal focused ultrasound ablation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月24日

查看全文
  • Mingzhen Shao,
  • Sumaiya Dipty,
  • Wei Zhang,
  • David Gillespie,
  • Matthew Zabriskie,
  • Ariana Page,
  • Randy Jensen,
  • Beatrice Kundsen,
  • Allison Payne,
  • Sarang Joshi,
  • Henrik Odéen

摘要

Glioblastoma multiforme (GBM) remains a highly lethal brain malignancy, and although maximal surgical resection can improve outcomes, many tumors are not amenable to aggressive surgery due to proximity to critical neurovascular structures and significant comorbidities. Transcranial focused ultrasound (tcFUS) is a non-invasive alternative which has shown promise in treating neurological diseases, such as treatment of Essential Tremor using thermal ablation. However, for neuro-oncology applications thermal ablations have been constrained by skull heating, long cooling intervals, and limited treatable volumes. Instead, microbubble-enabled FUS approaches, including blood brain barrier opening and drug delivery, have shown promise. Additionally, using microbubbles to non-thermally ablate tissue by inertial microbubble cavitation, causing hemorrhagic lesions, has been investigated pre-clinically. Non-thermal ablation (NTA) uses lower time-averaged powers than thermal ablation, avoiding skull heating issues and enlarging the treatment envelope, as well as lower FUS duty cycle so that larger tumor volumes can potentially be treated. In this work, we quantitatively evaluate the use of magnetic resonance imaging (MRI) for lesion assessment following phase-shift microbubble (PSMB)-enabled tcFUS-mediated NTA in rat model involving seven male Fischer rats, including five healthy rats and two rats bearing F98-LN glioma tumors. We first characterize hemorrhage-related hypointensities in standard T2*-weighted (T2*w) gradient recalled-echo MRI in healthy rat brain across a range of echo times (TE = 5.1–26.0 ms). Ground-truth hemorrhage is derived from registered hematoxylin and eosin (H&E) histology, enabling quantitative comparisons between MRI and histology. We identify a TE = 26 ms as yielding the highest spatial agreement with histology, as measured by soft Dice. To investigate improvements of the immediate outcome prediction, we further introduce a learning-based framework that leverages standard of care multi-parametric MRI (T1w and T2w, in addition to T2*w). Using a simple baseline model as proof of concept, we show that multi-parametric fusion (soft Dice 0.51) can predict histology-derived lesion/hemorrhage patterns 45% higher accuracy than T2*w imaging alone (soft Dice 0.35 using the TE=26 ms echo) in healthy brain. Finally, we demonstrate the robustness of the proposed analysis within the treatment region in a rat brain tumor model, supporting the translational potential of multiparametric MRI-based assessment for non-thermal tcFUS therapies.