G. Maggi

Hypofractionated Stereotactic Radiation Therapy in Recurrent High-Grade Glioma: A New Challenge

Purpose The aim of this study was to evaluate outcomes of hypofractionated stereotactic radiation therapy (HSRT) in patients re-treated for recurrent high-grade glioma. Materials and Methods From January 2006 to September 2013, 25 patients were treated. Six patients underwent radiation therapy alone, while 19 underwent combined treatment with surgery and/or chemotherapy. Only patients with Karnofsky Performance Status (KPS) > 70 and time from previous radiotherapy greater than 6 months were re-irradiated. The mean recurrent tumor volume was 35 cm3 (range, 2.46 to 116.7 cm3), and most of the patients (84%) were treated with a total dose of 25 Gy in five fractions (range, 20 to 50 Gy in 5-10 fractions). Results The median follow-up was 18 months (range, 4 to 36 months). The progression-free survival (PFS) at 1 and 2 years was 72% and 34% and the overall survival (OS) 76% and 50%, respectively. No severe toxicity was recorded. In univariate and multivariate analysis extent of resection at diagnosis significantly influenced PFS and OS (p < 0.01). Patients with smaller recurren tumor volume treated had better local control and survival. Indeed, the 2-year PFS was 40% (≤ 50 cm3) versus 11% (p=0.1) and the 2-year OS 56% versus 33% (> 50 cm3), respectively (p=0.26). Conclusion In our experience, HSRT could be a safe and feasible therapeutic option for recurrent high grade glioma even in patients with larger tumors. We believe that a multidisciplinary evaluation is mandatory to assure the best treatment for selected patients. Local treatment should also be considered as part of an integrated approach.

Outcome Evaluation of Oligometastatic Patients Treated with Surgical Resection Followed by Hypofractionated Stereotactic Radiosurgery (HSRS) on the Tumor Bed, for Single, Large Brain Metastases

Purpose The aim of this study was to evaluate the benefit of a combined treatment, surgery followed by adjuvant hypofractionated stereotactic radiosurgery (HSRS) on the tumor bed, in oligometastatic patients with single, large brain metastasis (BM). Methods and Materials Fom January 2011 to March 2015, 69 patients underwent complete surgical resection followed by HSRS with a total dose of 30Gy in 3 daily fractions. Clinical outcome was evaluated by neurological examination and MRI 2 months after radiotherapy and then every 3 months. Local progression was defined as radiographic increase of the enhancing abnormality in the irradiated volume, and brain distant progression as the presence of new brain metastases or leptomeningeal enhancement outside the irradiated volume. Surgical morbidity and radiation-therapy toxicity, local control (LC), brain distant progression (BDP), and overall survival (OS) were evaluated. Results The median preoperative volume and maximum diameter of BM was 18.5cm3 (range 4.1–64.2cm3) and 3.6cm (range 2.1-5-4cm); the median CTV was 29.0cm3 (range 4.1–203.1cm3) and median PTV was 55.2cm3 (range 17.2–282.9cm3). The median follow-up time was 24 months (range 4–33 months). The 1-and 2-year LC in site of treatment was 100%; the median, 1-and 2-year BDP was 11.9 months, 19.6% and 33.0%; the median, 1-and 2-year OS was 24 months (range 4–33 months), 91.3% and 73.0%. No severe postoperative morbidity or radiation therapy toxicity occurred in our series. Conclusions Multimodal approach, surgery followed by HSRS, can be an effective treatment option for selected patients with single, large brain metastases from different solid tumors.

A New Class of Wavelet-Based Metrics for Image Similarity Assessment

In this paper, we propose a new class of image similarity metrics based on a wavelet decomposition. By suitably combining weighted contributions of the different dyadic frequency bands, we define a class of similarity measures and we prove it is a metric. Moreover, we discuss the theoretical relationship between the novel class of metrics and the well-known structural similarity index (SSIM) and its multiscale versions (MSSSIM and CWSSIM). By using standard benchmark indexes over a reference database in the literature (the TID2013 database), we test the efficiency of the newly defined metrics in performing similarity assessment. We compare the performance of our metric with other well-known indexes in the literature, such as SSIM, FPH, MSSSIM, CWSSIM and PSNR, to demonstrate its improvement over the current state of the art, which becomes more evident when the query image is the one identified by the worst level of degradation which is perceived by the human visual system, as coded by the standard mean opinion score stored in the database.

Erratum to: A New Class of Wavelet-Based Metrics for Image Similarity Assessment

In this paper, we propose a new class of image similarity metrics based on a wavelet decomposition. By suitably combining weighted contributions of the different dyadic frequency bands, we define a class of similarity measures and we prove it is a metric. Moreover, we discuss the theoretical relationship between the novel class of metrics and the well-known structural similarity index (SSIM) and its multiscale versions (MSSSIM and CWSSIM). By using standard benchmark indexes over a reference database in the literature (the TID2013 database), we test the efficiency of the newly defined metrics in performing similarity assessment. We compare the performance of our metric with other well-known indexes in the literature, such as SSIM, FPH, MSSSIM, CWSSIM and PSNR, to demonstrate its improvement over the current state of the art, which becomes more evident when the query image is the one identified by the worst level of degradation which is perceived by the human visual system, as coded by the standard mean opinion score stored in the database. The original version of this article was revised: The double vertical bars are inserted instead of single vertical bars in Equation 9. B Silvia Bertoluzza silvia.bertoluzza@imati.cnr.it Maria Grazia Albanesi mariagrazia.albanesi@unipv.it 1 Department of Electrical, Computer and Biomedical Engineering, University of Pavia, Via Ferrata 1, 27100 Pavia, Italy 2 CNR IMATI Enrico Magenes, via Ferrata 1, 27100 Pavia, Italy 3 ICH Humanitas, Milan, Italy

OC-0077: Are pitch and roll compensations required in all pathologies? A data analysis of 2945 fractions

adjacent structures. For these patients a mask was created from the GTV by a 2cm expansion after which the GTV itself was removed (figure C,D), effectively registering the adjacent structures. This method was evaluated on five weekly fractions of 24 patients. The second method was applied on patients with a non-attached tumor. In this method the local rigid registration was expanded by a scaling factor such that the regressing tumor in the CBCT was magnified to the original size of the tumor of the reference CT-scan during the registration (figure G,H). This method was applied on 5 patients and also five weekly fractions were evaluated. Bland-Altman analysis was applied to quantify the limits of agreement between these registration methods and the clinically approved registrations. All automatic registrations were visually validated to assess the success rate. Results: The limits of agreement between the registration method for regressing tumors attached to surrounding structures showed limits of agreement with the clinical method of -2.6—2.9mm for the LR direction, -2.9—2.8mm for the CC direction and -3.1—3.2mm for the AP direction. The alignment differences between these two methods were 1.3 (LR), 1.4 (CC) and 1.4 mm (AP) systematically and 1.0, 1.1 and 1.2mm randomly. This automatic method had a success rate of 91%. The limits of agreement between the registration method for non-attached tumors and the clinical method were larger with -6.0—4.1mm (LR), -8.5—7.1mm (CC) and -3.3—4.3mm (AP). The alignment differences between these two methods were 4.0 (LR), 3.9 (CC) and 3.6mm (AP) systematically and 4.0, 3.3 and 2.4mm randomly. The success rate of these automatic registrations was 100%. Conclusions: The registration method developed for regressing tumors attached to surrounding structures proved to be a reliable method for automatic tumor registration. The registration method for regressing non-attached tumors is promising but needs further investigation on a larger patient cohort.

A Modular Registration Algorithm for Medical Images

The aim of this communication is to present the design of a code for image registration based on a modular structure which allows to easily combine and interchange different image models, transformation classes and image “distance” functionals, which are dealt with by independent modules which can be implemented transparently to each other. The code is tested by comparing several registration strategies on medical images.