· · · Repository hosted by TU Delft Library

The 3D fusion of tracked ultrasound with a diagnostic CT image has multiple benefits in a variety of interventional applications for oncology. Still, manual registration is a considerable drawback to theclinical workflow and hinders the widespread clinical adoption of this technique. In this paper, we propose a method to allow for an image-based automated registration, aligning multimodal images of the liver. We adopt a model-based approach that rigidly matches segmentedliver shapes from ultrasound (U/S) and diagnostic CT imaging. Towards this end, a novel method which combines a dynamic region-growingmethod with a graph-based segmentation framework is introduced to address the challenging problem of liver segmentation from U/S. The method is able to extract liver boundary from U/S images after a partial surface is generated near the principal vector from an electromagnetically tracked U/S liver sweep. The liver boundary is subsequently expanded by modeling the problem as a graph-cut minimization scheme, where cost functions used to detect optimal surface topology aredetermined from adaptive priors of neighboring surface points. Thisallows including boundaries affected by shadow areas by compensatingfor varying levels of contrast. The segmentation of the liver surface is performed in 3D space for increased accuracy and robustness. The method was evaluated in a study involving 8 patients undergoing biopsy or radiofrequency ablation of the liver, yielding promising surface segmentation results based on ground-truth comparison. The proposed extended segmentation technique improved the fiducial landmarkregistration error compared to a point-based registration (7.2mm vs. 10.2mm on average, respectively), while yielding a statistically insignificant differences in tumor target registration error (p > 0.05) compared to state-of-the-art methods.

Purpose Prostate cancer (PCA) is the second most frequent cause of cancer-related death in men in the United States and Europe. Transrectal ultrasound (TRUS) guided systematic prostate biopsy is the standard of care for detection and diagnosis of PCA. However, due to inadequate visualization of PCA in ultrasound, the false negative rate of systematic TRUS-guided biopsy is 15-30%. In this work, a novel prostate guidance system using image fusion of live TRUS with pre-acquired prostate magnetic resonance imaging (MRI) is presented and validated in pre-clinical and clinical studies. Methods (A) Fusion guidance system: Endorectal ultrasound probe (Philips C9-5, Andover, MA,USA) biopsy guides were equipped with electromagnetic (EM) trackingsensors. The pose of the sensors was calibrated relative to the image area of the probe, enabling realtime spatial tracking of the TRUSimages. A software application was created supporting registrationof the live ultrasound with pre-acquired MRI using the following steps. (1) The prostate in the 3-dimensional (3D) MRI image was segmented using a semi-automatic segmentation tool. (2) A spatial sweep (base to apex in transverse view) across the prostate with the trackedTRUS probe was reconstructed into a 3D TRUS volume and was segmentedautomatically using adaptive local shape statistics. (3) Registration between the TRUS volume and the MRI volume was initialized basedon the known sweep geometry and the TRUS and MRI segmentations. (4)The TRUS-MRI registration was optimized interactively by combining manual manipulation of individual rotational and translational degrees of freedom (DOFs) with subsequent automatic optimization of the remaining DOFs using the iterative closest point (ICP) algorithm. (5)After registration, the live 2D TRUS image was shown side-by-side with the corresponding multi-planar reconstruction (MPR) of the MRI image. The MRI-based segmentation and any MRI-identified points of interest (POIs) were also superimposed on the live images. This workflow is illustrated in Figure 1. (B) Validation: The system and workflow were validated in phantom studies (as reported earlier), dog models, and clinical studies. In 5 dog models, MRI-visible but ultrasound-occult targets were created by injecting 1/32” synthetic ruby balls, diluted Gadolinium, or diluted Feridex® into the dog prostate. The fusion system was used to inject a secondary ruby or Feridex® marker in the MRI-identified target location. The spatial distance between the primary and secondary injections in subsequent MRI were usedto define the overall spatial accuracy. Figure 2 shows T1-weighted MRI slices of one of the injected targets in a dog prostate (white arrow) before and after the secondary injection, and the targeted injection of the secondary fiducial (black arrow). (C) Clinical study: In 203 patients with elevated prostate-specific antigen (PSA) or abnormal digital rectal exam (DRE), multi-parametric MRI (T2-weighted,diffusion-weighted, dynamic contrast-enhanced, and magnetic resonance spectroscopy) of the prostate was obtained on a 3 Tesla Philips Achieva (Andover, MA, USA) using an endorectal coil (BPX-30; Medrad, Pittsburgh, Pa, USA). The MRI was read by 2 radiologists and lesionssuspicious for PCA were identified. The MRI lesions were categorizedinto low, moderate and high suspicion based on the number of MRI sequences positive for that lesion (1-2 sequences positive: low; 3: moderate; 4: high). Subsequently, the patients underwent systematic 12-core TRUS-guided biopsy and TRUS-MRI fusion-targeted biopsy of MRI-identified suspicious lesions. All patients provided written informed consent. 10 patients were inevaluable because non-standard equipment was used or because of other reasons. The positive biopsy rates for systematic biopsy, targeted biopsy and for the combined approach(systematic + targeted) were compared, and were correlated with theMRI suspicion labels. Results In 5 dog prostates, a total of 10 target markers (2 Feridex, 4 Gadolinium and 4 synthetic ruby balls) and10 secondary fiducials (2 ruby balls, 8 Feridex injections) were injected. All 10 of the targets and 9 of the fiducials could be identified in follow-up MRI. The mean +- standard deviation of the distance between targets and secondary injections was 5.0 +- 2.5 mm. The clinical patient population had a mean age of 61.5 years (median 61, range 40 – 82) and a mean PSA of 8.5 ng/ml (mean 5.8, range 0.0 – 103.0). 133 patients had a prior prostate biopsy, of which 75 were biopsy. Figure 3 shows the fusion display provided for biopsy of an MRI-identified “high suspicion” target in a 67 year old male. In the study population there was a significant increase in the per-patient and per-core positive biopsy rates with increasing MRI suspicion level, for systematic as well as for targeted and combined biopsies. Also, targeted positive core rates were significantly (p<0.01) higher than systematic core rates in patients with moderate or high MRI-suspicion but were equivalent in patients with low suspicion (see Figure4). For high suspicion patients, the targeted positive core rate (46.5%) was more than double the systematic positive core rate (22.4%).Furthermore, in the patient group with moderate or high suspicion,the combined approach had significantly (p<0.05) higher per-patientpositive biopsy rates than systematic biopsy alone. Conclusions A system enabling MRI-targeted prostate biopsy by fusing pre-acquired MRI with live TRUS outside the MRI gantry was developed, validated, and clinically tested. The spatial accuracy of the fusion system wassufficient to target clinically significant prostate cancer. MRI-based cancer suspicion categories correlated well with biopsy-proven cancer detection rates, suggesting an important role of MRI in prostate cancer management. MRI fusion targeting significantly increased cancer detection rates in select patient groups, and may allow improved out-of-gantry management of patients with moderate or high MRI suspicion.

In this paper, we present a novel approach to rigidly register intraoperative electromagnetically tracked ultrasound (US) with preoperative magnetic resonance (MR) images. The clinical rationale for thiswork is to allow accurate needle placement during thermal ablation of liver metastases using multimodal imaging. We adopt a model-basedapproach that rigidly matches segmented liver surface shapes obtained from the multimodal image volumes. Towards this end, a shape-constrained deformable surface model combining the strengths of both deformable and active shape models is used to segment the liver surfacefrom the MR scan. It incorporates a priori shape information while external forces guide the deformation and adapts the model to a target structure. The liver boundary is extracted from US by merging a dynamic region-growing method with a graph-based segmentation framework anchored on adaptive priors of neighboring surface points. Registration is performed with a weighted ICP algorithm with a physiological penalizing term. The MR segmentation model was trained with 30 datasets and validated on a separate cohort of 10 patients with corresponding ground truth. The accuracy and robustness of the method wereassessed by registering four US/MR datasets, yielding accurate landmark registration errors (3.7 +/- 0.69mm) and high robustness, and isthus acceptable for radiofrequency clinical applications.

Purpose: To evaluate the effect of different b-values on intravoxelincoherent motion (IVIM) and diffusion parameters for prostate cancer detection. Materials and methods: Thirty three patients (mean age of 61.6 years, mean serum PSA of 10 ng/dl) undergoing endorectal coil MRI of the prostate underwent multiparametric imaging includingdiffusion weighted (DW) imaging with five b-values (0, 188, 375, 563and 750 s/mm2), T2 weighting and dynamic contrast enhanced MRI. Diffusion coefficients were obtained from a simple mono-exponential fitusing different non-zero b-values. A simplified IVIM model was used to generate perfusion fractions, by combining both the measured and the extrapolated diffusion data at a b-value of zero. Correlationswere made with the results of DCE-MRI using an extended Tofts pharmacokinetic model. Pathologic correlation was obtained by precisely targeting the needle via a fused MRI-Transrectal Ultrasound (MR-TRUS)image-guided biopsy system. Results: Diffusion coefficients differentiated tumors from normal tissues in the prostate using all possible combinations of non-zero b-values; however, perfusion fractionsdemonstrated large variations depending on the choice of b-values.Exclusion of the highest b-value of 750 (s/mm2) led to better correlations of perfusion fraction with DCE-MRI and predicted the presenceof cancer independent of diffusion. Conclusions: Estimates of perfusion fraction using IVIM obtained on DW-MRI correlate with DCE-MRIparameters and are predictive for cancer in MRI of the prostate. Perfusion fraction therefore represents another independent parameterto help differentiate prostate cancers from surrounding benign tissue using multiparametric MRI