Cascade and Fusion: A Deep Learning Approach for Camouflaged Object Sensing

The demand for the sensor-based detection of camouflage objects widely exists in biological research, remote sensing, and military applications. However, the performance of traditional object detection algorithms is limited, as they are incapable of extracting informative parts from low signal-to-noise ratio features. To address this problem, we propose Camouflaged Object Detection with Cascade and Feedback Fusion (CODCEF), a deep learning framework based on an RGB optical sensor that leverages a cascaded structure with Feedback Partial Decoders (FPD) instead of a traditional encoder–decoder structure. Through a selective fusion strategy and feedback loop, FPD reduces the loss of information and the interference of noises in the process of feature interweaving. Furthermore, we introduce Pixel Perception Fusion (PPF) loss, which aims to pay more attention to local pixels that might become the edges of an object. Experimental results on an edge device show that CODCEF achieved competitive results compared with 10 state-of-the-art methods.

P294 AMT-101, a gut selective oral IL-10 fusion: results from a Phase 1b study in patients with active Ulcerative Colitis

Background IL-10 is a pleiotropic anti-inflammatory cytokine that restores mucosal homeostasis. Genetic deficiencies in IL-10 cause intestinal inflammation in mice and are associated with early onset IBD in humans. In mice, this can be reversed by addition of exogenous IL-10. Subcutaneous use of recombinant human IL-10 treatment in patients with Crohn’s Disease led to symptomatic and endoscopic improvement, but dose limiting systemic toxicity including anemia and thrombocytopenia was observed and appeared to be mediated by induction of gamma-interferon. AMT-101 is an oral, gut selective, fusion protein of IL-10 and a carrier protein that mediates transcytosis through intestinal enterocytes into the lamina propria (LP). Pre-clinical studies demonstrated that AMT-101 efficiently transits the epithelial cell barrier, targets mucosal immune cells in the LP, and results in improved colitis disease severity without systemic exposure. Methods A multiple ascending dose Ph1b POC trial was conducted in 16 patients with active UC where patients were randomized 3:1 to receive AMT-101 at doses of 1, 3, 10, or 30 mg or PBO, orally, once daily for 14 days. Primary and secondary outcome measures included safety and PK. Exploratory endpoints included changes in fecal calprotectin, histology, and stool microbiome. Results AMT-101 was safe and well tolerated; all AEs were mild to moderate and self limiting. No systemic toxicities associated with prior administration of IL-10 (e.g.anemia or thrombocytopenia) were observed. Systemic levels of AMT-101 were below limit of quantitation, given the GI restricted design. 1 mg and 3 mg AMT-101 led to placebo-adjusted mean reductions of FCP of 44% and 27%, respectively, in patients with baseline FCP>150 μg/g. Paired biopsy samples were centrally read in a blinded manner using the Geboes scoring system. Improvements were observed in 60% (6/10) of patients on AMT-101, compared with 0% (0/2) patients treated with placebo. Microbiome analyses revealed changes in phyla abundance in the 1mg and 3mg treatment cohorts (Bacteroidetes: 28.3% ± 31.4 to 51.3% ± 21.6; Firmicutes: 21.8% ± 12.7 to 30% ± 7.3), including Roseburia hominis and Faecalibacterium prausnitzii species. Conclusion The results of this Phase 1b study confirm that once daily, oral AMT-101 was safe and well tolerated without SAEs previously observed with systemic IL-10 and without systemic exposure, by design. Doses at 10 mg or less suggest potential clinical efficacy paired with an enhancement of favorable enteric commensal bacteria after only 14 days of treatment. These findings support AMT-101 as actively exerting an immunomodulatory effect in the intestinal lamina propria and support ongoing Ph2 trials of AMT-101.

Does Thoracic Kyphosis Have Any Importance in Selective Versus Nonselective Fusion Preference in Patients With Lenke Type 5C Adolescent Idiopathic Scoliosis? A Contribution to the Guidance of the Lenke Classification

BackgroundAlthough Lenke classification analyses the sagittal plane as (+), N, and (-), it does not consider it in the choice of treatment, and it has limitations with overall thoracic kyphosis (TK). To investigate the importance of TK for treatment preference in patients with Lenke 5C adolescent idiopathic scoliosis (AIS) by comparing radiological outcomes of the patients who underwent selective fusion (SF) or nonselective fusion (NSF). MethodsThirty-two patients with Lenke type 5C AIS were included and then divided into two groups as per the fusion procedure used in the surgical treatment. SF group including 17 patients (15 females; mean age = 16 years, age range, 14–21) with normal TK and NSF group including 15 patients (11 females; mean age = 17 years, age range, 13–26) with thoracic hyper-kyphosis. Thorocolumbar/lumbar (TL/L) Cobb, thoracic (T) Cobb, TK and lumbar lordosis (LL) were measured on standing spine radiographs preoperatively and at the final follow-up. The correction rates (CR) of each radiographic parameter were calculated.ResultsNo significant differences were observed in the mean CR of all radiographic parameters, except TK and LL correction rates. The mean CR of TK was significantly higher in NSF group (-17% [range, -100–69]) than in SF group (67% [range, 9–100]) (p = 0.000). Likewise, the mean CR of LL was found significantly higher in NSF group (12.47% [range, -100–51]) than in SF group (-2.41% [range, -75–47]) (p = 0.036).ConclusionIn patients in whom Lenke's sagittal modifier is N, SF can be performed efficiently. NSF should be preferred in those with Lenke's sagittal modifiers (+) as TK can be better controlled with NSF.Level of Evidence: 3