Convolutional Neural Network-based Reconstruction for Positronium Annihilation Localization

A hermetic novel detector composed of 200 Bismuth germanium oxide crystal scintillators and 393 channel silicon photomultipliers has been developed for positronium (Ps) annihilation study. This compact 4π detector is capable of simultaneously detecting γ-ray decay in all directions, enabling not only the study of visible and invisible exotic decay processes but also tumor localization in positron emission tomography for small animals. In this study, we investigate the use of a convolutional neural network (CNN) for the localization of the Ps annihilation synonymous with tumor localization. The 2-γ decay systems of the Ps annihilation from the 22Na and 18F radioactive sources are simulated using GEANT4. The simulated data sets are preprocessed by applying energy cuts. The spatial error in the XY plane from CNN is compared to that from the classical centroiding, weighted k-means algorithm. The feasibility of the CNN-based Ps an-nihilation reconstruction with tumor localization is discussed.

OS07.5A Diagnostic Instrument for Mild Aphasia (DIMA): sensitive and valuable addition to standard language assessment in glioma patients

BACKGROUND Low-grade glioma (LGG) patients typically suffer from mild aphasia that often cannot be detected with standard aphasia tests. The Diagnostic Instrument for Mild Aphasia (DIMA) is the first standardized test-battery to assess mild language disorders. We investigate pre- and postoperative linguistic abilities of LGG and high-grade glioma (HGG) patients with the DIMA. METHODS The DIMA consists of subtests that tap phonology (word, compound, non-word, sentence repetition), semantics (odd-picture-out), and syntax (sentence completion). Additionally, we administered the Boston Naming Test, Category- and Letter Fluency, and the Token Test. Patients were assessed before awake surgery (T1, N=98), three-months (T2, N=69), and one-year (T3, N=30) postoperatively. DIMA performance was compared to healthy controls (N=214). Group differences were examined with parametric (t-test) and nonparametric (Mann-Whitney-U, Wilcoxon) tests. RESULTS DIMA: Preoperatively, patients deviated on sentence repetition and sentence completion (p<0.05). HGG patients performed worse than LGG on word, non-word, and sentence repetition (p<0.05). There was no effect of hemispheric tumor localization. At T2, compound repetition and odd-picture-out also became impaired (p<0.05) and there was a decline compared to T1 on all repetition tasks (p<0.05). At T3, only sentence completion remained impaired (p<0.01) with a deterioration compared to T1 (p<0.01). Standard tests: At T1, patients were impaired on BNT, Category- and Letter Fluency (p<0.01). HGG patients performed worse than LGG patients on BNT and TT (p<0.01). Patients with left-hemispheric tumors performed worse on BNT and Letter Fluency compared to right-hemispheric tumors (p<0.05). At T2, TT also became impaired (p<0.05) and patients declined compared to T1 on Verbal Fluency tests (p<0.01). At T3, only BNT and Category Fluency remained impaired (p<0.05), with no significant declines compared to T1. CONCLUSION The DIMA is the first test-battery to detect peri-operative impairments at different linguistic levels (phonology, semantics, syntax) in patients with left- or right-hemispheric gliomas. It even appeared more sensitive to detect surgical effects than standard tests: all phonological DIMA subtests captured short-term decline (T1-T2), in line with earlier evidence for the value of (non-)word repetition. DIMA sentence completion detected long-term decline (T1-T3), reflecting earlier spontaneous speech analyses. As expected, Verbal Fluency was also sensitive to short-term postoperative decline. Left-hemispheric tumor localization only affected standard test performance. HGG patients had more severe impairments than LGG on DIMA repetition and standard tests (BNT and TT). We advise adding the DIMA to standard language evaluation of glioma patients, as it allows for more detailed counseling about language outcome.