Optical-readout time projection chambers (TPCs) produce megapixel-scale images whose fine- grained topological information is essential for rare-event searches, but whose size challenges real- time data selection. We present an unsupervised, reconstruction-based anomaly-detection strategy for fast region-of-interest (ROI) extraction that operates directly on minimally processed cam- era frames. A convolutional autoencoder (AE) trained exclusively on pedestal images learns the detector noise morphology without labels, simulation, or fine-grained calibration. Applied to standard data-taking frames, localized reconstruction residuals identify particle-induced struc- tures, from which compact ROIs are extracted via thresholding and spatial clustering. Using real data from the CYGNO optical TPC prototype, we compare two pedestal-trained AE configurations that differ only in their training objective, enabling a controlled study of its impact. The best con- figuration retains (93.0 ± 0.2)% of reconstructed signal intensity while discarding (97.8 ± 0.1)% of the image area, with an inference time of ∼25 ms per frame on a consumer GPU. The results demonstrate that careful design of the training objective is critical for effective reconstruction- based anomaly detection and that pedestal-trained AEs provide a transparent and detector- agnostic baseline for online data reduction in optical TPCs
Fast reconstruction-based ROI triggering via anomaly detection in the CYGNO optical TPC / Amaro, F D; Antonietti, R; Baracchini, Elisabetta; Benussi, Luigi; Capoccia, Cesidio; Caponero, Michele Arturo; Carvalho, L G M De; Cavoto, G; Costa, I A; Croce, A; D'Astolfo, M; D'Imperio, Giulia; Dho, Giorgio; Di Marco, E; Dos Santos, J M F; Fiorina, Davide; Iacoangeli, F; Islam, Zahoor Ul; Kemp, Ernesto; Lima Jr, Herman P; Maccarrone, Giovanni; Mano, R D P; Gaspar Marques, David José; Mazzitelli, Giovanni; Meloni, Pietro; Messina, A; Monno, V; Monteiro, C M B; Nobrega, R A; Oppedisano, G M; Pains, Igor Fonseca; Paoletti, E; Petrucci, Fabrizio; Piacentini, Stefano; Pierluigi, D; Pinci, Davide; Renga, F; Russo, A; Saviano, G; O C Silva, P A; Spooner, N J; Tesauro, Roberto; Tomassini, Sandro; Tozzi, Donatella. - In: MACHINE LEARNING: SCIENCE AND TECHNOLOGY. - ISSN 2632-2153. - 7:2(2026). [10.1088/2632-2153/ae5c5b]
Fast reconstruction-based ROI triggering via anomaly detection in the CYGNO optical TPC
Cavoto, GMembro del Collaboration Group
;Messina, AMembro del Collaboration Group
;Monno, VMembro del Collaboration Group
;Oppedisano, G MMembro del Collaboration Group
;Saviano, G;Tozzi, Donatella
2026
Abstract
Optical-readout time projection chambers (TPCs) produce megapixel-scale images whose fine- grained topological information is essential for rare-event searches, but whose size challenges real- time data selection. We present an unsupervised, reconstruction-based anomaly-detection strategy for fast region-of-interest (ROI) extraction that operates directly on minimally processed cam- era frames. A convolutional autoencoder (AE) trained exclusively on pedestal images learns the detector noise morphology without labels, simulation, or fine-grained calibration. Applied to standard data-taking frames, localized reconstruction residuals identify particle-induced struc- tures, from which compact ROIs are extracted via thresholding and spatial clustering. Using real data from the CYGNO optical TPC prototype, we compare two pedestal-trained AE configurations that differ only in their training objective, enabling a controlled study of its impact. The best con- figuration retains (93.0 ± 0.2)% of reconstructed signal intensity while discarding (97.8 ± 0.1)% of the image area, with an inference time of ∼25 ms per frame on a consumer GPU. The results demonstrate that careful design of the training objective is critical for effective reconstruction- based anomaly detection and that pedestal-trained AEs provide a transparent and detector- agnostic baseline for online data reduction in optical TPCsI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
