RESUMO
Due to scarcity of anomaly situations in the early manufacturing stage, an unsupervised anomaly detection (UAD) approach is widely adopted which only uses normal samples for training. This approach is based on the assumption that the trained UAD model will accurately reconstruct normal patterns but struggles with unseen anomalies. To enhance the UAD performance, reconstruction-by-inpainting based methods have recently been investigated, especially on the masking strategy of suspected defective regions. However, there are still issues to overcome: (1) time-consuming inference due to multiple masking, (2) output inconsistency by random masking, and (3) inaccurate reconstruction of normal patterns for large masked areas. Motivated by this, this study proposes a novel reconstruction-by-inpainting method, dubbed Excision And Recovery (EAR), that features single deterministic masking based on the ImageNet pre-trained DINO-ViT and visual obfuscation for hint-providing. Experimental results on the MVTec AD dataset show that deterministic masking by pre-trained attention effectively cuts out suspected defective regions and resolves the aforementioned issues 1 and 2. Also, hint-providing by mosaicing proves to enhance the performance than emptying those regions by binary masking, thereby overcomes issue 3. The proposed approach achieves a high performance without any change of the model structure. Promising results are shown through laboratory tests with public industrial datasets. To suggest EAR be possibly adopted in various industries as a practically deployable solution, future steps include evaluating its applicability in relevant manufacturing environments.
RESUMO
BACKGROUND: Margin status is one of the most significant prognostic factors after curative surgery for middle bile duct (MBD) cancer. Bile duct resection (BDR) is commonly converted to pancreaticoduodenectomy (PD) to achieve R0 resection. Additionally, adjuvant treatment is actively performed after surgery to improve survival. However, the wider the range of surgery, the higher the chance of complications; this, in turn, makes adjuvant treatment impossible. Nevertheless, no definitive surgical strategy considers the possible complication rates and subsequent adjuvant treatment. We aimed to investigate the appropriate surgical type considering the margin status, complications, and adjuvant treatment in MBD cancer. MATERIALS AND METHODS: From 2008 to 2017, 520 patients diagnosed with MBD cancer at the Samsung Medical Center were analyzed retrospectively according to the operation type, margin status, complications, and adjuvant treatment. The R1 group was defined as having a carcinoma margin. RESULTS: The 5-year survival rate for patients who underwent R0 and R1 resection was 54.4% and 33.3%, respectively (p = 0.131). Prognostic factors affecting the overall survival were the age, preoperative CA19-9 level, T stage, and N stage, but not the operation type, margin status, complications, or adjuvant treatment. The complication rates were 11.5% and 29.8% in the BDR and PD groups, respectively (p < 0.001). We observed no significant difference in the adjuvant treatment ratio according to complications (p = 0.675). Patients with PD who underwent R0 resection and could not undergo chemotherapy because of complications reported better survival rates than those with BDR who underwent R1 resection after adjuvant treatment (p = 0.003). CONCLUSION: The survival outcome of patients with R1 margins who underwent BDR did not match those with R0 margins after PD, even after adjuvant treatment. Due to improvements in surgical techniques and the ability to resolve complications, surgical complications exert a marginal effect on survival. Therefore, surgeons should secure R0 margins to achieve the best survival outcomes.