The Effect of Slice Thickness on Contours of Brain Metastases for Stereotactic Radiosurgery

Deep learning for the harmonization of structural MRI scans: a survey

Medical imaging datasets for research are frequently collected from multiple imaging centers using different scanners, protocols, and settings. These variations affect data consistency and compatibility across different sources. Image harmonization is a critical step to mitigate the effects of factors like inherent differences between various vendors, hardware upgrades, protocol changes, and scanner calibration drift, as well as to ensure consistent data for medical image processing techniques. Given the critical importance and widespread relevance of this issue, a vast array of image harmonization methodologies have emerged, with deep learning-based approaches driving substantial advancements in recent times. The goal of this review paper is to examine the latest deep learning techniques employed for image harmonization by analyzing cutting-edge architectural approaches in the field of medical image harmonization, evaluating both their strengths and limitations. This paper begins by providing a comprehensive fundamental overview of image harmonization strategies, covering three critical aspects: established imaging datasets, commonly used evaluation metrics, and characteristics of different scanners. Subsequently, this paper analyzes recent structural MRI (Magnetic Resonance Imaging) harmonization techniques based on network architecture, network learning algorithm, network supervision strategy, and network output. The underlying architectures include U-Net, Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), flow-based generative models, transformer-based approaches, as well as custom-designed network architectures. This paper investigates the effectiveness of Disentangled Representation Learning (DRL) as a pivotal learning algorithm in harmonization. Lastly, the review highlights the primary limitations in harmonization techniques, specifically the lack of comprehensive quantitative comparisons across different methods. The overall aim of this review is to serve as a guide for researchers and practitioners to select appropriate architectures based on their specific conditions and requirements. It also aims to foster discussions around ongoing challenges in the field and shed light on promising future research directions with the potential for significant advancements.

Comparison between postcontrast thin-slice T1-weighted 2D spin echo and 3D T1-weighted SPACE sequences in the detection of brain metastases at 1.5 and 3 T

OBJECTIVES

Accurate detection of metastatic brain lesions (MBL) is critical due to advances in radiosurgery. We compared the results of three readers in detecting MBL using T1-weighted 2D spin echo (SE) and sampling perfection with application-optimized contrasts using different flip angle evolution (SPACE) sequences with whole-brain coverage at both 1.5 T and 3 T.

METHODS

Fifty-six patients evaluated for MBL were included and underwent a standard protocol (1.5 T, n = 37; 3 T, n = 19), including postcontrast T1-weighted SE and SPACE. The rating was performed by three raters in two sessions > six weeks apart. The true number of MBL was determined using all available imaging including follow-up. Intraclass correlations for intra-rater and inter-rater agreement were calculated. Signal intensity ratios (SIR; enhancing lesion, white matter) were determined on a subset of 46 MBL > 4 mm. A paired t-test was used to evaluate postcontrast sequence order and SIR. Reader accuracy was evaluated by the coefficient of determination.

RESULTS

A total of 135 MBL were identified (mean/subject 2.41, SD 6.4). The intra-rater agreement was excellent for all 3 raters (ICC = 0.97-0.992), as was the inter-rater agreement (ICC = 0.995 SE, 0.99 SPACE). Subjective qualitative ratings were lower for SE images; however, signal intensity ratios were higher in SE sequences. Accuracy was high in all readers for both SE (R2 0.95-0.96) and SPACE (R2 0.91-0.96) sequences.

CONCLUSIONS

Although SE sequences are superior to gradient echo sequences in the detection of small MBL, they have long acquisition times and frequent artifacts. We show that T1-weighted SPACE is not inferior to standard thin-slice SE sequences in the detection of MBL at both imaging fields.

CRITICAL RELEVANCE STATEMENT

Our results show the suitability of 3D T1-weighted turbo spin echo (TSE) sequences (SPACE, CUBE, VISTA) in the detection of brain metastases at both 1.5 T and 3 T.

KEY POINTS

• Accurate detection of brain metastases is critical due to advances in radiosurgery. • T1-weighted SE sequences are superior to gradient echo in detecting small metastases. • T1-weighted 3D-TSE sequences may achieve high resolution and relative insensitivity to artifacts. • T1-weighted 3D-TSE sequences have been recommended in imaging brain metastases at 3 T. • We found T1-weighted 3D-TSE equivalent to thin-slice SE at 1.5 T and 3 T.

Preoperative growth dynamics of untreated glioblastoma: Description of an exponential growth type, correlating factors, and association with postoperative survival

Background

Little is known about the growth dynamics of untreated glioblastoma and its possible influence on postoperative survival. Our aim was to analyze a possible association of preoperative growth dynamics with postoperative survival.

Methods

We performed a retrospective analysis of all adult patients surgically treated for newly diagnosed glioblastoma at our center between 2010 and 2020. By volumetric analysis of data of patients with availability of ≥3 preoperative sequential MRI, a growth pattern was aimed to be identified. Main inclusion criterion for further analysis was the availability of two preoperative MRI scans with a slice thickness of 1 mm, at least 7 days apart. Individual growth rates were calculated. Association with overall survival (OS) was examined by multivariable.

Results

Out of 749 patients screened, 13 had ≥3 preoperative MRI, 70 had 2 MRI and met the inclusion criteria. A curve estimation regression model showed the best fit for exponential tumor growth. Median tumor volume doubling time (VDT) was 31 days, median specific growth rate (SGR) was 2.2% growth per day. SGR showed negative correlation with tumor size (rho = -0.59, P < .001). Growth rates were dichotomized according to the median SGR.OS was significantly longer in the group with slow growth (log-rank: P = .010). Slower preoperative growth was independently associated with longer overall survival in a multivariable Cox regression model for patients after tumor resection.

Conclusions

Especially small lesions suggestive of glioblastoma showed exponential tumor growth with variable growth rates and a median VDT of 31 days. SGR was significantly associated with OS in patients with tumor resection in our sample.

Measurement of brainstem diameter in small-breed dogs using magnetic resonance imaging

Measurement of brainstem diameters (midbrain, pons, and medulla oblongata)is of potential clinical significance, as changes in brainstem size may decrease or increase due to age, neurodegenerative disorders, or neoplasms. In human medicine, numerous studies have reported the normal reference range of brainstem size, which is hitherto unexplored in veterinary medicine, particularly for small-breed dogs. Therefore, this study aims to investigate the reference range of brainstem diameters in small-breed dogs and to correlate the measurements with age, body weight (BW), and body condition score (BCS). Herein, magnetic resonance (MR) images of 544 small-breed dogs were evaluated. Based on the exclusion criteria, 193 dogs were included in the midbrain and pons evaluation, and of these, 119 dogs were included in the medulla oblongata evaluation. Using MR images, the height and width of the midbrain, pons, and medulla oblongata were measured on the median and transverse plane on the T1-weighted image. For the medulla oblongata, two points were measured for each height and width. The mean values of midbrain height (MH), midbrain width (MW), pons height (PH), pons width (PW), medulla oblongata height at the fourth ventricle level (MOHV), medulla oblongata height at the cervicomedullary (CM) junction level (MOHC), rostral medulla oblongata width (RMOW), and caudal medulla oblongata width (CMOW) were 7.18 ± 0.56 mm, 17.42 ± 1.21 mm, 9.73 ± 0.64 mm, 17.23 ± 1.21 mm, 6.06 ± 0.53 mm, 5.77 ± 0.40 mm, 18.93 ± 1.25 mm, and 10.12 ± 1.08 mm, respectively. No significant differences were found between male and female dogs for all the measurements. A negative correlation was found between age and midbrain diameter, including MH (p < 0.001) and MW (p  = 0.002). All brainstem diameters were correlated positively with BW (p  < 0.05). No significant correlation was found between BCS and all brainstem diameters. Brainstem diameters differed significantly between breeds (p  < 0.05), except for MW (p  = 0.137). This study assessed linear measurements of the brainstem diameter in small-breed dogs. We suggest that these results could be useful in assessing abnormal conditions of the brainstem in small-breed dogs.

Optimizing imaging resolution in brain MRI: understanding the impact of technical factors

Magnetic resonance imaging (MRI) exams are essential for diagnostic procedures, but their lengthy duration and associated costs limit their accessibility. Shorter scan times would reduce expenses and allow for more MRI exams, expanding the range of diagnostic procedures. This study investigated technical factors that could decrease scan time without compromising image quality, including field-of-view (FOV), phase field of view, phase oversampling, cross-talk, brain MRI imaging resolution, and scan time. Data were collected from September 2021 to June 2022. All patients underwent brain scans in the transverse plane following a standardized protocol using a 1.5-tesla Siemens Avanto MRI scanner. The protocol employed T2-weighted Turbo Spin Echo imaging. Twenty-four cases were included in this study. Initially, all participants underwent brain MRI scans using the original protocols with axial sections. The results indicated that altering the FOV phase and phase oversampling significantly affected the scan time, whereas other factors did not have a direct impact. The original protocol had a scan time of 3.47 minutes with a FOV of 230 mm, 90% FOV phase, and 0% phase oversampling. After implementing the modified protocol, the scan time was reduced to 2.18 minutes with a FOV of 217 mm and 93.98% phase oversampling of 13.96%. Statistical analysis confirmed the high significance of FOV phase and phase oversampling in reducing scan time. By optimizing these technical factors, MRI exams can be performed more efficiently, resulting in shorter scan times and potentially reducing costs. This would enhance patient comfort and enable a greater number of MRI exams, facilitating a more comprehensive range of diagnostic procedures.

DeSeg: auto detector-based segmentation for brain metastases

Delineation of brain metastases (BMs) is a paramount step in stereotactic radiosurgery treatment. Clinical practice has specific expectation on BM auto-delineation that the method is supposed to avoid missing of small lesions and yield accurate contours for large lesions. In this study, we propose a novel coarse-to-fine framework, named detector-based segmentation (DeSeg), to incorporate object-level detection into pixel-wise segmentation so as to meet the clinical demand. DeSeg consists of three components: a center-point-guided single-shot detector to localize the potential lesion regions, a multi-head U-Net segmentation model to refine contours, and a data cascade unit to connect both tasks smoothly. Performance on tiny lesions is measured by the object-based sensitivity and positive predictive value (PPV), while that on large lesions is quantified by dice similarity coefficient (DSC), average symmetric surface distance (ASSD) and 95% Hausdorff distance (HD95). Besides, computational complexity is also considered to study the potential of method in real-time processing. This study retrospectively collected 240 BM patients with Gadolinium injected contrast-enhanced T1-weighted magnetic resonance imaging (T1c-MRI), which were randomly split into training, validating and testing datasets (192, 24 and 24 scans, respectively). The lesions in the testing dataset were further divided into two groups based on the volume size (smallS: ≤1.5 cc,N= 88; largeL: > 1.5 cc,N= 15). On average, DeSeg yielded a sensitivity of 0.91 and a PPV of 0.77 on S group, and a DSC of 0.86, an ASSD 0f 0.76 mm and a HD95 of 2.31 mm onLgroup. The results indicated that DeSeg achieved leading sensitivity and PPV for tiny lesions as well as segmentation metrics for large ones. After our clinical validation, DeSeg showed competitive segmentation performance while kept faster processing speed comparing with existing 3D models.

Protocol for mathematical prediction of patient response and survival to immune checkpoint inhibitor immunotherapy

This protocol describes the application of a mechanistic mathematical model of immune checkpoint inhibitor (ICI) immunotherapy to patient tumor imaging data for predicting solid tumor response and patient survival under ICI intervention. We describe steps for data collection and processing, data pipelines, and approaches to increase precision. The protocol is highly predictive as early as the first restaging after treatment start and can be used with standard-of-care imaging measures. For complete details on the use and execution of this protocol, please refer to Butner et al. (2020)¹ and Butner et al. (2021).².

Quantitative Relaxometry Metrics for Brain Metastases Compared to Normal Tissues: A Pilot MR Fingerprinting Study

The purpose of the present pilot study was to estimate T1 and T2 metric values derived simultaneously from a new, rapid Magnetic Resonance Fingerprinting (MRF) technique, as well as to assess their ability to characterize-brain metastases (BM) and normal-appearing brain tissues. Fourteen patients with BM underwent MRI, including prototype MRF, on a 3T scanner. In total, 108 measurements were analyzed: 42 from solid parts of BM's (21 each on T1 and T2 maps) and 66 from normal-appearing brain tissue (11 ROIs each on T1 and T2 maps for gray matter [GM], white matter [WM], and cerebrospinal fluid [CSF]). The BM's mean T1 and T2 values differed significantly from normal-appearing WM (p < 0.05). The mean T1 values from normal-appearing GM, WM, and CSF regions were 1205 ms, 840 ms, and 4233 ms, respectively. The mean T2 values were 108 ms, 78 ms, and 442 ms, respectively. The mean T1 and T2 values for untreated BM (n = 4) were 2035 ms and 168 ms, respectively. For treated BM (n = 17) the T1 and T2 values were 2163 ms and 141 ms, respectively. MRF technique appears to be a promising and rapid quantitative method for the characterization of free water content and tumor morphology in BMs.

Dedicated isotropic 3-D T1 SPACE sequence imaging for radiosurgery planning improves brain metastases detection and reduces the risk of intracranial relapse

BACKGROUND

Stereotactic radiosurgery (SRS) is increasingly used for brain metastases (BM) patients, but distant intracranial failure (DIF) remains the principal disadvantage of this focal therapeutic approach. The objective of this study was to determine if dedicated SRS imaging would improve lesion detection and reduce DIF.

METHODS

Between 02/2020 and 01/2021, SRS patients at a tertiary care institution underwent dedicated treatment planning MRIs of the brain including MPRAGE and SPACE post-contrast sequences. DIF was calculated using the Kaplan-Meier method; comparisons were made to a historical consecutive cohort treated using MPRAGE alone (02/2019-01/2020).

RESULTS

134 patients underwent 171 SRS courses for 821 BM imaged with both MPRAGE and SPACE (primary cohort). MPRAGE sequence evaluation alone detected 679 lesions. With neuroradiologists evaluating SPACE and MPRAGE, an additional 108 lesions were identified (p<0.001). Upon multidisciplinary review, 34 additional lesions were identified. Compared to the historical cohort (103 patients, 135 SRS courses, 479 BM), the primary cohort had improved median time to DIF (13.5 vs. 5.1 months, p=0.004). The benefit was even more pronounced for patients treated for their first SRS course (18.4 vs. 6.3 months, p=0.001). SRS using MPRAGE and SPACE was associated with a 60% reduction in risk of DIF compared to the historical cohort (HR: 0.40; 95%CI: 0.28-0.57, p<0.001).

CONCLUSIONS

Among BM patients treated with SRS, a treatment planning SPACE sequence in addition to MPRAGE substantially improved lesion detection and was associated with a statistically significant and clinically meaningful prolongation in time to DIF, especially for patients undergoing their first SRS course.

Radiosurgery for Five to Fifteen Brain Metastases: A Single Centre Experience and a Review of the Literature

Purpose

Stereotactic radiosurgery (SRS) is now mainstream for patients with 1-4 brain metastases however the management of patients with 5 or more brain metastases remains controversial. Our aim was to evaluate the clinical outcomes of patients with 5 or more brain metastases and to compare with published series as a benchmarking exercise.

Methods

Patients with 5 or more brain metastases treated with a single isocentre dynamic conformal arc technique on a radiosurgery linac were identified from the institutional database. Endpoints were local control, distant brain failure, leptomeningeal disease and overall survival. Dosimetric data were extracted from the radiosurgery plans. Series reporting outcomes following SRS for multiple brain metastases were identified by a literature search.

Results

36 patients, of whom 35 could be evaluated, received SRS for 5 or more brain metastases between February 2015 and October 2021. 25 patients had 5-9 brain metastases (group 1) and 10 patients had 10-15 brain metastases (group 2). The mean number of brain metastases in group 1 was 6.3 (5-9) and 12.3 (10-15) in group 2. The median cumulative irradiated volume was 4.6 cm³ (1.25-11.01) in group 1 and 7.2 cm³ (2.6-11.1) in group 2. Median follow-up was 12 months. At last follow-up, local control rates per BM were 100% and 99.8% as compared with a median of 87% at 1 year in published series. Distant brain failure was 36% and 50% at a median interval of 5.2 months and 7.4 months after SRS in groups 1 and 2 respectively and brain metastasis velocity at 1 year was similar in both groups (9.7 and 11). 8/25 patients received further SRS and 7/35 patients received whole brain radiotherapy. Median overall survival was 10 months in group 1 and 15.7 months in group 2, which compares well with the 7.5 months derived from the literature. There was one neurological death in group 2, leptomeningeal disease was rare (2/35) and there were no cases of radionecrosis.

Conclusion

With careful patient selection, overall survival following SRS for multiple brain metastases is determined by the course of the extracranial disease. SRS is an efficacious and safe modality that can achieve intracranial disease control and should be offered to patients with 5 or more brain metastases and a constellation of good prognostic factors.

The Promise of Magnetic Resonance Imaging in Radiation Oncology Practice in the Management of Brain, Prostate, and GI Malignancies

Magnetic resonance imaging (MRI) has a key role to play at multiple steps of the radiotherapy (RT) treatment planning and delivery process. Development of high-precision RT techniques such as intensity-modulated RT, stereotactic ablative RT, and particle beam therapy has enabled oncologists to escalate RT dose to the target while restricting doses to organs at risk (OAR). MRI plays a critical role in target volume delineation in various disease sites, thus ensuring that these high-precision techniques can be safely implemented. Accurate identification of gross disease has also enabled selective dose escalation as a means to widen the therapeutic index. Morphological and functional MRI sequences have also facilitated an understanding of temporal changes in target volumes and OAR during a course of RT, allowing for midtreatment volumetric and biological adaptation. The latest advancement in linear accelerator technology has led to the incorporation of an MRI scanner in the treatment unit. MRI-guided RT provides the opportunity for MRI-only workflow along with online adaptation for either target or OAR or both. MRI plays a key role in post-treatment response evaluation and is an important tool for guiding decision making. In this review, we briefly discuss the RT-related applications of MRI in the management of brain, prostate, and GI malignancies.

Synthetic MRI for Radiotherapy Planning for Brain and Prostate Cancers: Phantom Validation and Patient Evaluation

Purpose

We aimed to evaluate the accuracy of T₁ and T₂ mappings derived from a multispectral pulse sequence (magnetic resonance image compilation, MAGiC^®) on 1.5-T MRI and with conventional sequences [gradient echo with variable flip angle (GRE-VFA) and multi-echo spin echo (ME-SE)] compared to the reference values for the purpose of radiotherapy treatment planning.

Methods

The accuracy of T₁ and T₂ measurements was evaluated with 2 coils [head and neck unit (HNU) and BODY coils] on phantoms using descriptive statistics and Bland–Altman analysis. The reproducibility and repeatability of T₁ and T₂ measurements were performed on 15 sessions with the HNU coil. The T₁ and T₂ synthetic sequences obtained by both methods were evaluated according to quality assurance (QA) requirements for radiotherapy. T₁ and T₂in vivo measurements of the brain or prostate tissues of two groups of five subjects were also compared.

Results

The phantom results showed good agreement (mean bias, 8.4%) between the two measurement methods for T₁ values between 490 and 2,385 ms and T₂ values between 25 and 400 ms. MAGiC^® gave discordant results for T₁ values below 220 ms (bias with the reference values, from 38% to 1,620%). T₂ measurements were accurately estimated below 400 ms (mean bias, 8.5%) by both methods. The QA assessments are in agreement with the recommendations of imaging for contouring purposes for radiotherapy planning. On patient data of the brain and prostate, the measurements of T₁ and T₂ by the two quantitative MRI (qMRI) methods were comparable (max difference, <7%).

Conclusion

This study shows that the accuracy, reproducibility, and repeatability of the multispectral pulse sequence (MAGiC^®) were compatible with its use for radiotherapy treatment planning in a range of values corresponding to soft tissues. Even validated for brain imaging, MAGiC^® could potentially be used for prostate qMRI.

Convolutional Neural Network-aided Tuber Segmentation in Tuberous Sclerosis Complex Patients Correlates with EEG

OBJECTIVE

One of the clinical hallmarks of tuberous sclerosis complex is radiologically-identified cortical tubers present in most patients. Intractable epilepsy may require surgery, often involving invasive diagnostic procedures such as intracranial EEG. Identifying the location of the dominant tuber responsible for generating epileptic activities, is a critical issue. However, the link between cortical tubers and epileptogenesis is poorly understood. Given this, we hypothesized that tuber voxel intensity may be an indicator of the dominant epileptogenic tuber. Also, via tuber segmentation based on deep learning, we explore whether an automatic quantification of the tuber burden is feasible.

METHODS

We annotated tubers from structural MRIs across 29 TSC subjects, summarized tuber statistics in eight brain lobes, and determined suspected epileptogenic lobes from the same group using EEG monitoring data. Then logistic regression analyses are performed to demonstrate the linkage between the statistics of cortical tuber and the epileptogenic zones. Furthermore, we test the ability of a neural network to identify and quantify tuber burden.

RESULTS

Logistic regression analyses show that the volume and count of tubers per lobe, not the mean or variance of tuber voxel intensity, are positively correlated with electrophysiological data. In 47.6% of subjects, the lobe with the largest tuber volume concurred with the epileptic brain activity. A neural network model on the test dataset shows a sensitivity of 0.83 for localizing individual tubers. The predicted masks from the model highly correlated with the neurologist labels, thus may be a useful tool for determining tuber burden and searching for epileptogenic zone.

SIGNIFICANCE

we prove the feasibility of an automatic segmentation of tubers and a derivation of tuber burden across brain lobes. Our method may provide crucial insights in the treatment and outcome of TSC patients.

Target localization accuracy in frame‐based stereotactic radiosurgery: Comparison between MR‐only and MR/CT co‐registration approaches

Purpose

In frame‐based Gamma Knife (GK) stereotactic radiosurgery two treatment planning workflows are commonly employed; one based solely on magnetic resonance (MR) images and the other based on magnetic resonance/computed tomography (MR/CT) co‐registered images. In both workflows, target localization accuracy (TLA) can be deteriorated due to MR‐related geometric distortions and/or MR/CT co‐registration uncertainties. In this study, the overall TLA following both clinical workflows is evaluated for cases of multiple brain metastases.

Methods

A polymer gel‐filled head phantom, having the Leksell stereotactic headframe attached, was CT‐imaged and irradiated by a GK Perfexion unit. A total of 26 4‐mm shots were delivered at 26 locations directly defined in the Leksell stereotactic space (LSS), inducing adequate contrast in corresponding T2‐weighted (T2w) MR images. Prescribed shot coordinates served as reference locations. An additional MR scan was acquired to implement the “mean image” distortion correction technique. The TLA for each workflow was assessed by comparing the radiation‐induced target locations, identified in MR images, with corresponding reference locations. Using T1w MR and CT images of 15 patients (totaling 81 lesions), TLA in clinical cases was similarly assessed, considering MR‐corrected data as reference. For the MR/CT workflow, both global and region of interest (ROI)‐based MR/CT registration approaches were studied.

Results

In phantom measurements, the MR‐corrected workflow demonstrated unsurpassed TLA (median offset of 0.2 mm) which deteriorated for MR‐only and MR/CT workflows (median offsets of 0.8 and 0.6 mm, respectively). In real‐patient cases, the MR‐only workflow resulted in offsets that exhibit a significant positive correlation with the distance from the MR isocenter, reaching 1.1 mm (median 0.6 mm). Comparable results were obtained for the MR/CT‐global workflow, although a maximum offset of 1.4 mm was detected. TLA was improved with the MR/CT‐ROI workflow resulting in median/maximum offsets of 0.4 mm/1.1 mm.

Conclusions

Subpixel TLA is achievable in all workflows. For the MR/CT workflow, a ROI‐based MR/CT co‐registration approach could considerably increase TLA and should be preferred instead of a global registration.

Advances in the management of breast cancer brain metastases

The development of breast cancer (BC) brain metastases (BrM) is a common complication of advanced disease, occurring in up to half of the patients with advanced disease depending on the subtype. The management of BCBrM requires complex multidisciplinary care including local therapy, surgical resection and/or radiotherapy, palliative care, and carefully selected systemic therapies. Significant progress has been made in the human epidermal growth factor receptor 2-positive (HER2+) BCBrM population due to novel brain penetrable systemic therapies. Increased inclusion of patients with BCBrM in clinical trials using brain-penetrant systemic therapies recently led to the first FDA approval of a HER2-directed therapy specifically in the BCBrM population in the last year. Advances for the treatment of HR+/HER2- and TNBC BCBrM subgroups continue to evolve. In this review, we will discuss the diagnosis and multidisciplinary care of BCBrM. We focus on recent advances in neurosurgery, radiation therapy, and systemic treatment therapies with intracranial activity. We also provide an overview of the current clinical trial landscape for patients with BCBrM.