2025 Theses Doctoral

Development and optimization of a clinical harmonic motion imaging system for focused ultrasound ablation monitoring and assessment

Li, Xiaoyue Judy

Breast cancer is the most commonly diagnosed invasive cancer in women both globally, in which breast cancer accounts for 25% of new cancer cases, and within the United States, with an estimated 31% of new cancer diagnoses [1, 2, 3]. Benign (non-cancerous) breast disease has a much larger prevalence; it is estimated that 80% of breast biopsies are benign [4, 5, 6].

Of all breast cancer cases, more than 90% are non-metastatic at the time of diagnosis [7]. In non-metastatic patients, the therapeutic goal is tumor eradication and recurrence prevention. The standard of care consists of systemic therapy, determined by molecular subtype, and local therapy, which consists of either total mastectomy or excision plus radiation [7]. Metastatic breast cancer is also treated systemically according to subtype, with goals of prolonging life and palliating symptoms. Local therapy is typically only used for palliative purposes in patients with metastatic disease [7]. In benign cases, surgical exicision may be performed primarily for symptom palliation [8, 9, 10, 4].

Minimally invasive and non-invasive techniques for tumor extirpation are being investigated as an alternative to surgical excision, for both benign and malignant neoplasms. Minimally invasive and non-invasive procedures are advantageous in economic benefits, reduced recovery time, absence of general anesthesia, and improved cosmetic result. Additionally, in conventional tumor surgical excisions, there is a lack of intraoperative imaging that would aid in achieving clear margins, resulting in high rates of reoperation to excise residual tumor [11, 12]. Incorporating intraoperative imaging techniques could enable higher rates of clear margins with breast conserving procedures. Such monitoring methods may be easier to achieve with minimally invasive and non-invasive techniques than with surgery.

Ultrasound elastography guides FUS ablation by monitoring changing mechanical properties of tissue that occur during thermal ablation. Harmonic Motion Imaging (HMI) is an elastography method, in which an oscillatory radiation force is used to generate tissue displacement at a specified frequency. As displacement is measured on the axis of the applied radiation force, HMI is advantageous for FUS monitoring at the site of the FUS transducer focus. Due to excitation of displacement at a specified frequency, HMI is advantageous in distinguishing displacement information from extraneous noise. HMI has previously been shown to distinguish stiffness related changes during FUS ablation [13, 14, 15, 16, 17, 18].In this dissertation, we aim to further improve our HMI guided FUS (HMIgFUS) system, and test its efficacy in pre-clinical studies and, for the first time, in in vivo clinical studies.

First, we explored two methods of optimizing the HMIgFUS system. In conventional HMIgFUS, imaging is performed concurrently with FUS application without interruption, such that FUS interference must be removed from radiofrequency (RF) data prior to displacement estimation. However, under high FUS pressure and long FUS treatment durations, previously used methods for FUS interference filtering, namely notch filtering, were ineffective. Thus, my first sub-aim was to investigate methods of FUS interference filtering, for the purpose of optimizing FUS ablation efficiency (high FUS power output) while minimizing FUS interference (for accurate displace- ment estimation). To achieve this goal, I performed HMIgFUS in an ex vivo canine liver, with a combination of FUS durations and peak positive pressures. I evaluated the performance of three methods of FUS interference filtering on these datasets: notch filtering, FUS-net, and interleaved HMIgFUS. We found that the interleaved method for HMIgFUS was found to be significantly robust in avoiding FUS interference in all tested cases for FUS ablation monitoring, especially cases with high FUS pressures and long durations, as opposed to traditional notch filtering and FUS-net filtering. FUS-net and notch filtering are advantageous in that they can be used with continuous FUS output, and thus may be more advantageous for low-intensity FUS purposes. FUS-net has also exhibited greater potential in FUS interference filtering while preserving frequency information, as compared with notch filtering.

In the second sub-aim, we investigated the use of multiple amplitude modulated (AM) frequency HMIgFUS for improved lesion size estimation. In previous studies, amplitude modulation was performed during HMIgFUS with a sinusoidal signal at a single frequency [19]. However, previous studies have shown benefits of multiple AM-frequency HMI imaging for improved contrast and CNR [20, 21, 22]. Multi-AM-frequency HMI was first performed in a tissue-mimicking phantom with embedded inclusions at three different sizes and stiffnesses, to evaluate the relationship of AM-frequency on inclusion size estimation. Multi-AM-frequency HMIgFUS ablation was then performed and evaluated in two ex vivo chicken muscle specimens. We found that low AM-frequency HMIgFUS may be ideal for lesion size estimation. In chicken muscle HMIgFUS, the lowest AM frequency was found to correlate the best with the final lesion size, as determined with gross pathology. However, contrast over time was highest and varied more over time with high AM frequencies, suggesting that high AM frequencies may be better suited for lesion progression monitoring.

In our second aim, preclinical studies of HMIgFUS in a breast cancer mouse model were performed in order to test our methods prior to clinical application, as well as to measure longitudinal responses to FUS ablation monotherapy. The breast cancer mouse model model was studied first longitudinally without treatment, acquiring HMI displacement images at the primary mammary tumor and at the liver. This study was performed to familiarize us with the mouse model and to assess HMI as a longitudinal monitoring imaging method in untreated mice. We found that HMI can distinguish differences in stiffness between the primary tumor and neighboring tissue, and the liver and its neighboring tissue. There also appears to be stiffening in both the primary tumor and liver with disease progression, which in the case of the liver, may be correlated to metastasis progression. Two preclinical HMIgFUS studies were then performed; one study in which mice developed two tumor sites, so that each mouse served as its own control; and one in which mice developed one tumor site each, so that longitudinal development of the single tumor following FUS ablation could be assessed. In both studies, acute and longitudinal effects of FUS ablation were assessed, and the monitoring and predictive ability of HMI during FUS was also assessed. We found a significant correlation between multi-AM HMIgFUS displacement percent change and 3D HMI displacement imaging percent change, suggesting that multi-AM HMIgFUS can reliably estimate the ablation lesion size during ablation. Our ablation results were further validated with bioluminescence imaging, which showed decreased luciferin expression in treated mice immediately following ablation. Finally, treated mice experienced a slower rate of tumor growth compared to untreated mice, suggesting the clinical utility of FUS ablation.

Finally, HMIgFUS was demonstrated for the first time in clinic, and the efficacy of HMIgFUS ablation and monitoring was assessed. In this first HMIgFUS clinical study, patients with non-metastatic early stage breast cancer or benign breast tumors that were scheduled by their clinical care team to receive surgical resection of the breast tumor were enrolled. We performed HMIgFUS under anesthesia in the operating room immediately prior to tumor resection. Among the ten patients out of 12 in which HMIgFUS monitoring was found to acquire and estimate reliable displacement data, we found a mean HMI displacement change of -53.3%±16.9, comparing the maximum displacement acquired during HMIgFUS with the displacement estimated at the final HMIgFUS timepoint. These initial results suggest that HMI displacement can measure FUS induced mechanical changes in tumor tissue, and that the measured displacement may differ among patients.