Reflective filters are characterized by a frequency response with good matching at the band of interest and usually reactive impedance out of those frequencies which may adversely affect the system performance. On the other hand, reflectionless filters are characterized by good matching characteristic not only at the interest frequencies, but in the whole frequency spectrum which improves the overall linearity, efficiency, and reduces instability scenarios at the system level. Although several reflectionless structures can be found in the literature, the concatenation of different reflectionless sections, combined with the use of acoustic resonators has not been exploited yet. The particular electrical behavior of acoustic wave resonators, where two different resonant frequencies are found, allow to obtain a frequency response with high selectivity due to the presence of transmission zeros below and above the passband. A bandpass filter has been designed following the described procedure with a fractional bandwidth F BW = 2%, a pair of transmission zeros below and above the bandpass, and an improved out-of-band rejection with respect conventional topologies.

A blind source separation method (cNMF) was used to extract characteristic metabolic patterns from PRESS MRSI 3T acquired from areas of contrast enhancement in a retrospective set of 31 glioblastoma patients, one month after the end of concomitant chemoradiotherapy with temozolomide. The aim was to evaluate whether these patterns were predictive of true progression or pseudoprogression. They were used as input for supervised classifiers, achieving a maximum of 81% balanced accuracy. A moderate association between extracted patterns and outcome was detected by Cramer’s V. Spatial source distribution with nosologic maps points to MRSI-detected metabolic heterogeneity as cause for classifiers’ performance.

In vivo magnetic resonance spectroscopy (MRS) has two modalities, single-voxel (SV) and multivoxel (MV), in which one or more contiguous grids of SVs are acquired. Purpose: To test whether MV grids can be classified with models trained with SV. Methods: Retrospective study. Training dataset: Multicenter multiformat SV INTERPRET, 1.5T. Testing dataset: MV eTumour, 3T. Two classification tasks were completed: 3-class (meningioma vs. aggressive vs. normal) and 4-class (meningioma vs. low-grade glioma vs. aggressive vs. normal). Five different methods were tested for feature selection. The classification was implemented using linear discriminant analysis (LDA), random forest, and support vector machines. The evaluation was completed with balanced error rate (BER) and area under the curve (AUC) on both sets. The accuracy in class prediction was calculated by developing a solid tumor index (STI) and segmentation accuracy with the Dice score. Results: The best method was sequential forward feature selection combined with LDA, with AUCs = 0.95 (meningioma), 0.89 (aggressive), 0.82 (low-grade glioma), and 0.82 (normal). STI was 66% (4-class task) and 71% (3-class task) because two cases failed completely and two more had suboptimal STI as defined by us. Discussion: The reasons for failure in the classification of the MV test set were related to the presence of artifacts.