Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

oncology Oncology

Sometimes it is better to just make it simple. De-escalation of oncoplastic and reconstructive procedures.

In Breast (Edinburgh, Scotland)

Simple breast conservation surgery (sBCS) has technically advanced onto oncoplastic breast procedures (OBP) to avoid mastectomy and improve breast cancer patients' psychosocial well-being and cosmetic outcome. Although OBP are time-consuming and expensive, we are witnessing an increase in their use, even for cases that could be managed with sBCS. The choice between keeping it simple or opting for more complex oncoplastic procedures is difficult. This review proposes a pragmatic approach in assisting this decision. Medical literature suggests that OBP and sBCS might be similar regarding local recurrence and overall survival, and patients seem to have higher satisfaction levels with the aesthetic outcome of OBP when compared to sBCS. However, the lack of comprehensive high-quality research assessing their safety, efficacy, and patient-reported outcomes hinders these supposed conclusions. Postoperative complications after OBP may delay the initiation of adjuvant RT. In addition, precise displacement of the breast volume is not effectively recorded despite surgical clips placement, making accurate dose delivery tricky for radiation oncologists, and WBRT preferable to APBI in complex OBP cases. With a critical eye on financial toxicity, patient satisfaction, and oncological outcomes, OBP must be carefully integrated into clinical practice. The thoughtful provision of informed consent is essential for decision-making between sBCS and OBP. As we look into the future, machine learning and artificial intelligence can potentially help patients and doctors avoid postoperative regrets by setting realistic aesthetic expectations.

Bonci E-A, Anacleto J Correia, Cardoso M-J

2023-Mar-13

Breast cancer, Breast conserving therapy, Oncoplastic surgery, Reconstructive breast surgery, de-escalation therapy

General General

Deep learning based MRI reconstruction with transformer.

In Computer methods and programs in biomedicine

Magnetic resonance imaging (MRI) has become one of the most powerful imaging techniques in medical diagnosis, yet the prolonged scanning time becomes a bottleneck for application. Reconstruction methods based on compress sensing (CS) have made progress in reducing this cost by acquiring fewer points in the k-space. Traditional CS methods impose restrictions from different sparse domains to regularize the optimization that always requires balancing time with accuracy. Neural network techniques enable learning a better prior from sample pairs and generating the results in an analytic way. In this paper, we propose a deep learning based reconstruction method to restore high-quality MRI images from undersampled k-space data in an end-to-end style. Unlike prior literature adopting convolutional neural networks (CNN), advanced Swin Transformer is used as the backbone of our work, which proved to be powerful in extracting deep features of the image. In addition, we combined the k-space consistency in the output and further improved the quality. We compared our models with several reconstruction methods and variants, and the experiment results proved that our model achieves the best results in samples at low sampling rates. The source code of KTMR could be acquired at https://github.com/BITwzl/KTMR.

Wu Zhengliang, Liao Weibin, Yan Chao, Zhao Mangsuo, Liu Guowen, Ma Ning, Li Xuesong

2023-Mar-01

Compress sensing, Deep learning, Magnetic resonance imaging (MRI), Transformer

Surgery Surgery

102 AI-Based Molecular Classification of Diffuse Gliomas using Rapid, Label-Free Optical Imaging.

In Neurosurgery ; h5-index 55.0

INTRODUCTION : Molecular classification has transformed the management of brain tumors by enabling more accurate prognostication and personalized treatment. Access to timely molecular diagnostic testing for brain tumor patients is limited, complicating surgical and adjuvant treatment and obstructing clinical trial enrollment.

METHODS : By combining stimulated Raman histology (SRH), a rapid, label-free, non-consumptive, optical imaging method, and deep learning-based image classification, we are able to predict the molecular genetic features used by the World Health Organization (WHO) to define the adult-type diffuse glioma taxonomy, including IDH-1/2, 1p19q-codeletion, and ATRX loss. We developed a multimodal deep neural network training strategy that uses both SRH images and large-scale, public diffuse glioma genomic data (i.e. TCGA, CGGA, etc.) in order to achieve optimal molecular classification performance.

RESULTS : One institution was used for model training (University of Michigan) and four institutions (NYU, UCSF, Medical University of Vienna, and University Hospital Cologne) were included for patient enrollment in the prospective testing cohort. Using our system, called DeepGlioma, we achieved an average molecular genetic classification accuracy of 93.2% and identified the correct diffuse glioma molecular subgroup with 91.5% accuracy within 2 minutes in the operating room. DeepGlioma outperformed conventional IDH1-R132H immunohistochemistry (94.2% versus 91.4% accuracy) as a first-line molecular diagnostic screening method for diffuse gliomas and can detect canonical and non-canonical IDH mutations.

CONCLUSIONS : Our results demonstrate how artificial intelligence and optical histology can be used to provide a rapid and scalable alternative to wet lab methods for the molecular diagnosis of brain tumor patients during surgery.

Hollon Todd Charles, Golfinos John G, Orringer Daniel A, Berger Mitchel, Hervey-Jumper Shawn L, Muraszko Karin M, Freudiger Christian, Heth Jason, Sagher Oren, Jiang Cheng, Chowdury Asadur, Moin Mustafa Nasir, Kondepudi Akhil, Aabedi Alexander Arash, Adapa Arjun R, Al-Holou Wajd, Wadiura Lisa, Widhalm Georg, Neuschmelting Volker, Reinecke David, Camelo-Piragua Sandra

2023-Apr-01

Radiology Radiology

Novel urinary protein panels for the non-invasive diagnosis of nonalcoholic fatty liver disease and fibrosis stages.

In Liver international : official journal of the International Association for the Study of the Liver

BACKGROUND & AIMS : There is an unmet clinical need for non-invasive tests to diagnose nonalcoholic fatty liver disease (NAFLD) and individual fibrosis stages. We aimed to test whether urine protein panels could be used to identify NAFLD, NAFLD with fibrosis (stage F≥1), and NAFLD with significant fibrosis (stage F≥2).

METHODS : We collected urine samples from 100 patients with biopsy-confirmed NAFLD and 40 healthy volunteers and proteomics and bioinformatics analyses were performed in this derivation cohort. Diagnostic models were developed for detecting NAFLD (UPNAFLD model), NAFLD with fibrosis (UPfibrosis model), or NAFLD with significant fibrosis (UPsignificant fibrosis model). Subsequently, the derivation cohort was divided into training and testing sets to evaluate the efficacy of these diagnostic models. Finally, in a separate independent validation cohort of 100 patients with biopsy-confirmed NAFLD and 45 healthy controls, urinary enzyme-linked immunosorbent assay analyses were undertaken to validate the accuracy of these newly diagnostic models.

RESULTS : The UPfibrosis model and the UPsignificant fibrosis model showed an AUROC of 0.863 (95% CI: 0.725-1.000) and 0.858 (95% CI: 0.712-1.000) in the training set; and 0.837 (95% CI: 0.711-0.963) and 0.916 (95% CI: 0.825-1.000) in the testing set, respectively. The UPNAFLD model showed excellent diagnostic performance and the area under the receiver operator characteristic curve (AUROC) exceeded 0.90 in the derivation cohort. In the independent validation cohort, the AUROC for all three of the above diagnostic models exceeded 0.80.

CONCLUSIONS : Our newly developed models constructed from urine protein biomarkers have good accuracy for non-invasively diagnosing liver fibrosis in NAFLD.

Feng Gong, Zhang Xiaoxun, Zhang Liangjun, Liu Wen-Yue, Geng Shi, Yuan Hai-Yang, Sha Jun-Cheng, Wang Xiao-Dong, Sun Dan-Qin, Targher Giovanni, Byrne Christopher D, Zheng Tian-Lei, Ye Feng, Zheng Ming-Hua, Chai Jin

2023-Mar-16

Diagnosis, Fibrosis, Liver biopsy, NAFLD, Urinary proteomics

General General

Revealing influencing factors on global waste distribution via deep-learning based dumpsite detection from satellite imagery.

In Nature communications ; h5-index 260.0

With the advancement of global civilisation, monitoring and managing dumpsites have become essential parts of environmental governance in various countries. Dumpsite locations are difficult to obtain in a timely manner by local government agencies and environmental groups. The World Bank shows that governments need to spend massive labour and economic costs to collect illegal dumpsites to implement management. Here we show that applying novel deep convolutional networks to high-resolution satellite images can provide an effective, efficient, and low-cost method to detect dumpsites. In sampled areas of 28 cities around the world, our model detects nearly 1000 dumpsites that appeared around 2021. This approach reduces the investigation time by more than 96.8% compared with the manual method. With this novel and powerful methodology, it is now capable of analysing the relationship between dumpsites and various social attributes on a global scale, temporally and spatially.

Sun Xian, Yin Dongshuo, Qin Fei, Yu Hongfeng, Lu Wanxuan, Yao Fanglong, He Qibin, Huang Xingliang, Yan Zhiyuan, Wang Peijin, Deng Chubo, Liu Nayu, Yang Yiran, Liang Wei, Wang Ruiping, Wang Cheng, Yokoya Naoto, Hänsch Ronny, Fu Kun

2023-Mar-15

Public Health Public Health

CCTCOVID: COVID-19 detection from chest X-ray images using Compact Convolutional Transformers.

In Frontiers in public health

COVID-19 is a novel virus that attacks the upper respiratory tract and the lungs. Its person-to-person transmissibility is considerably rapid and this has caused serious problems in approximately every facet of individuals' lives. While some infected individuals may remain completely asymptomatic, others have been frequently witnessed to have mild to severe symptoms. In addition to this, thousands of death cases around the globe indicated that detecting COVID-19 is an urgent demand in the communities. Practically, this is prominently done with the help of screening medical images such as Computed Tomography (CT) and X-ray images. However, the cumbersome clinical procedures and a large number of daily cases have imposed great challenges on medical practitioners. Deep Learning-based approaches have demonstrated a profound potential in a wide range of medical tasks. As a result, we introduce a transformer-based method for automatically detecting COVID-19 from X-ray images using Compact Convolutional Transformers (CCT). Our extensive experiments prove the efficacy of the proposed method with an accuracy of 99.22% which outperforms the previous works.

Marefat Abdolreza, Marefat Mahdieh, Hassannataj Joloudari Javad, Nematollahi Mohammad Ali, Lashgari Reza

2023

COVID-19, Compact Convolutional Transformers, Convolutional Neural Networks, deep learning, vision transformers