Search

Despite the success of convolutional neural networks (CNNs) in numerous computer vision tasks and their extraordinary generalization performances, several attempts to predict the generalization errors of CNNs have only been limited to a posteriori analyses thus far. A priori theories explaining the generalization performances of deep neural networks have mostly ignored the convolutionality aspect and do not specify why CNNs are able to seemingly overcome curse of dimensionality on computer vision tasks like image classification where the image dimensions are in thousands. Our work attempts to explain the generalization performance of CNNs on image classification under the hypothesis that CNNs operate on the domain of image patches. Ours is the first work we are aware of to derive an a priori error bound for the generalization error of CNNs and we present both quantitative and qualitative evidences in the support of our theory. Our patch-based theory also offers explanation for why data augmentation techniques like Cutout, CutMix and random cropping are effective in improving the generalization error of CNNs.

Tuberculosis (TB), caused by Mycobacterium TB, is one of the leading causes of major illness and death worldwide. Although cutaneous TB constitutes a small proportion of extrapulmonary TB burden, awareness regarding varied clinical presentations is necessary as early diagnosis and treatment are keys to reduce the morbidity of the disease. Here, we present a series of four rare cases of cutaneous TB presenting as vulval elephantiasis, Poncet's disease, disseminated TB, and anterior chest wall TB. These cases posed diagnostic difficulty due to unusual presentations. However, all the cases responded well to antitubercular treatment.

Background: Lumbar spine surgery is associated with significant postoperative pain. Interfascial plane blocks, such as erector spinae plane (ESP) and thoracolumbar interfascial plane (TLIP) blocks, can play a significant role in multimodal analgesic regimens., Methods: Sixty patients aged 18 to 60 years undergoing elective single or double-level lumbar discectomy or primary lumbar laminoplasty were recruited into this randomized double‑blind study. All patients received general anesthesia and were randomly allocated to either modified TLIP (mTLIP) block (group M) or ESP block (group E). Postoperative and intraoperative fentanyl consumption, and postoperative pain scores, were recorded., Results: Total 48 h postoperative fentanyl consumption was higher in Group M (189.66±141.11 µg) than in Group E (124.16±80.83 µg; P =0.031). In the first 24 postoperative hours, fentanyl consumption was higher in Group M (150.3±120.9 µg) than in group E (89.9±65.3 µg; P =0.01) but was similar between groups in postoperative hours 24to 48 (39.0±20.2 µg versus 34.7±17.1 µg in group M and group E, respectively; P =0.37). Additional intraoperative fentanyl requirement was 57.66±21.76 µg in group M compared with 40.33±21.89 µg in group E ( P <0.01). Postoperative pain scores were higher in group M than in group E at 1, 2, 4, 6, 12, and 24 hours postoperatively ( P <0.001), but similar at 48 hours ( P =0.164)., Conclusion: Compared with the mTLIP block, the ESP block was associated with lower pain scores and a small decrease in perioperative fentanyl consumption in patients undergoing lumbar spine surgeries. Both blocks could form a part of a multimodal analgesic regimen in spine surgery patients., Competing Interests: The authors have no conflicts of interest to disclose., (Copyright © 2023 Wolters Kluwer Health, Inc. All rights reserved.)

Introduction: Ultrasound (US)-guided interscalene (IS) block is a commonly performed block for shoulder and humerus surgery. Though it provides excellent analgesia, it is associated with hemidiaphragmatic paralysis and dyspnoea. Superior trunk (ST) block has been described, wherein the local anaesthetic is deposited around the ST block (formed by fusion of C5 and C6 nerve roots). This study aimed to determine whether ST block provides similar analgesic efficacy with lower incidence of diaphragmatic paresis in patients undergoing proximal humerus surgery., Material and Methods: A total of 62 patients scheduled to undergo unilateral internal fixation (plating) for proximal or mid shaft humerus fracture were randomised to 2 groups. Patients in group I received US-guided ST block while those in group II received US-guided IS block. Both groups received 15 mL of 0.5% bupivacaine. Diaphragmatic excursion was noted at baseline and after 30 minutes after the block. Postoperatively, the numerical rating scale score and requirement of opioids were documented., Results: The incidence of complete/incomplete paresis was statistically significantly lower in the ST group. Thirty eight percent of the patients (11) had complete paresis in the IS group, compared to none in the ST group. Partial paresis was observed in 62% of patients in the IS block group and 19% in the ST block group ( P < 0.001). The percentage reduction of movement was significantly higher in the IS group vs. the ST group ( P < 0.001). There was no difference in pain scores or the amount of opioid consumption between groups., Conclusions: ST block provides similar analgesia to IS block for proximal/mid humerus surgery with better preservation of diaphragmatic function. This could be a viable alternative in patients with compromised respiratory functions scheduled for such surgery.

Background and Aims: Hip replacement surgery is a commonly performed surgery with the aim of improving mobility in patients suffering from hip conditions. Though the modified suprainguinal approach of fascia iliaca block (SFIB) is commonly used, the analgesic efficacy is moderate and is associated with quadriceps weakness. The pericapsular nerve group (PENG) block has been used to block the sensory articular branches of the hip joint in various hip surgeries. This study aimed to compare SFIB with PENG block in terms of pain relief, opioid consumption and their adverse effects in patients undergoing primary total hip arthroplasties. (THA)., Methods: Seventy ASA I/II patients undergoing primary THA were enrolled in this double-blinded, randomized trial. Patients were randomly allocated to one of the two groups: Group P: ultrasound (US)-guided PENG block and Group S: patients received the US-guided SFIB., Results: Postoperatively, there was statistically significant difference in numerical rating scale (NRS) scores at all-time intervals. Total morphine consumption in 24 hours and 48 hours was statistically more in SFIB group. Five patients had quadriceps weakness in the SFIB group. There was no difference in any other adverse effects., Conclusion: US-guided PENG block significantly reduces perioperative morphine consumption and pain scores in THA patients when compared to SFI block. It is not associated with quadriceps weakness as seen in SFIB., Competing Interests: There are no conflicts of interest., (Copyright: © 2023 Indian Journal of Anaesthesia.)

Competing Interests: Nil.