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

Radiology Radiology

Deep learning to estimate RECIST in patients with NSCLC treated with PD-1 blockade.

In Cancer discovery ; h5-index 105.0

Real-world evidence (RWE) - conclusions derived from analysis of patients not treated in clinical trials - is increasingly recognized as an opportunity for discovery, to reduce disparities, and to contribute to regulatory approval. Maximal value of RWE may be facilitated through machine learning techniques to integrate and interrogate large and otherwise underutilized data sets. In cancer research, an ongoing challenge for RWE is the lack of reliable, reproducible, scalable assessment of treatment-specific outcomes. We hypothesized a deep learning model could be trained to use radiology text reports to estimate gold-standard Response Evaluation Criteria in Solid Tumors (RECIST)-defined outcomes. Using text reports from patients with non-small cell lung cancer treated with PD-1 blockade in a training cohort and two test cohorts, we developed a deep learning model to accurately estimate best overall response and progression-free survival. Our model may be a tool to determine outcomes at scale, enabling analyses of large clinical databases.

Arbour Kathryn C, Luu Anh Tuan, Luo Jia, Rizvi Hira, Plodkowski Andrew J, Sakhi Mustafa, Huang Kevin B, Digumarthy Subba R, Ginsberg Michelle S, Girshman Jeffrey, Kris Mark G, Riely Gregory J, Yala Adam, Gainor Justin F, Barzilay Regina, Hellmann Matthew D


Radiology Radiology

Atherosclerotic Plaque Burden on Abdominal CT: Automated Assessment With Deep Learning on Noncontrast and Contrast-enhanced Scans.

In Academic radiology

BACKGROUND : Abdominal aortic atherosclerotic plaque burden may have clinical significance but manual measurement is time-consuming and impractical.

PURPOSE : To perform external validation on an automated atherosclerotic plaque detector for noncontrast and postcontrast abdominal CT.

MATERIALS AND METHODS : The training data consisted of 114 noncontrast CT scans and 23 postcontrast CT urography scans. The testing data set consisted of 922 CT colonography (CTC) scans, and 1207 paired noncontrast and postcontrast CT scans from renal donors from a second institution. Reference standard data included manual plaque segmentations in the 137 training scans and manual plaque burden measurements in the 922 CTC scans. The total Agatston score and group (0-3) was determined using fully-automated deep learning software. Performance was assessed by measures of agreement, linear regression, and paired evaluations.

RESULTS : On CTC scans, automated Agatston scoring correlated highly with manual assessment (R2 = 0.94). On paired renal donor CT scans, automated Agatston scoring on postcontrast CT correlated highly with noncontrast CT (R2 = 0.95). When plaque burden was expressed as a group score, there was excellent agreement for both the CTC (weighted kappa 0.80 ± 0.01 [95% confidence interval: 0.78-0.83]) and renal donor (0.83 ± 0.02 [0.79-0.86]) assessments.

CONCLUSION : Fully automated detection, segmentation, and scoring of abdominal aortic atherosclerotic plaques on both pre- and post-contrast CT was validated and may have application for population-based studies.

Summers Ronald M, Elton Daniel C, Lee Sungwon, Zhu Yingying, Liu Jiamin, Bagheri Mohammedhadi, Sandfort Veit, Grayson Peter C, Mehta Nehal N, Pinto Peter A, Linehan W Marston, Perez Alberto A, Graffy Peter M, O’Connor Stacy D, Pickhardt Perry J


3D-UNet, Agatston score, Image processing, machine learning

General General

Trends and targets of various types of stem cell derived transfusable RBC substitution therapy: Obstacles that need to be converted to opportunity.

In Transfusion and apheresis science : official journal of the World Apheresis Association : official journal of the European Society for Haemapheresis

A shortage of blood during the pandemic outbreak of COVID-19 is a typical example in which the maintenance of a safe and adequate blood supply becomes difficult and highly demanding. So far, human RBCs have been produced in vitro using diverse sources: hematopoietic stem cells (SCs), embryonic SCs and induced pluripotent SCs. The existing, even safest core of conventional cellular bioproducts destined for transfusion have some shortcoming in respects to: donor -dependency variability in terms of hematological /immunological and process/ storage period issues. SCs-derived transfusable RBC bioproducts, as one blood group type for all, were highly complex to work out. Moreover, the strategies for their successful production are often dependent upon the right selection of starting source materials and the composition and the stability of the right expansion media and the strict compliance to GMP regulatory processes. In this mini-review we highlight some model studies, which showed that the efficiency and the functionality of RBCs that could be produced by the various types of SCs, in relation to the in-vitro culture procedures are such that they may, potentially, be used at an industrial level. However, all cultured products do not have an unlimited life due to the critical metabolic pathways or the metabolites produced. New bioreactors are needed to remove these shortcomings and the development of a new mouse model is required. Modern clinical trials based on the employment of regenerative medicine approaches in combination with novel large-scale bioengineering tools, could overcome the current obstacles in artificial RBC substitution, possibly allowing an efficient RBC industrial production.

Lanza Francesxco, Seghatchian Jerard


Artificial intelligence, Bioreactors, Expansion media, GMP regulatory processes, Induced pluripotent stem cell, Stem cells, Transfusable RBC bioproducts

Ophthalmology Ophthalmology

The role of neuroinflammation in the pathogenesis of glaucoma neurodegeneration.

In Progress in brain research

The chapter is a review enclosed in the volume "Glaucoma: A pancitopatia of the retina and beyond." No cure exists for glaucoma. Knowledge on the molecular and cellular alterations underlying glaucoma neurodegeneration (GL-ND) includes innovative and path-breaking research on neuroinflammation and neuroprotection. A series of events involving immune response (IR), oxidative stress and gene expression are occurring during the glaucoma course. Uveitic glaucoma (UG) is a prevalent acute/chronic complication, in the setting of chronic anterior chamber inflammation. Managing the disease requires a team approach to guarantee better results for eyes and vision. Advances in biomedicine/biotechnology are driving a tremendous revolution in ophthalmology and ophthalmic research. New diagnostic and imaging modalities, constantly refined, enable outstanding criteria for delimiting glaucomatous neurodegeneration. Moreover, biotherapies that may modulate or inhibit the IR must be considered among the first-line for glaucoma neuroprotection. This review offers the readers useful and practical information on the latest updates in this regard.

Pinazo-Durán Maria D, Muñoz-Negrete Francisco J, Sanz-González Silvia M, Benítez-Del-Castillo Javier, Giménez-Gómez Rafael, Valero-Velló Mar, Zanón-Moreno Vicente, García-Medina José J


Artificial intelligence, Glaucoma, Imaging, Neurodegeneration, Neuroinflammation, Neuroprotection, Uveitic glaucoma

General General

Adoption of Machine Learning in Intelligent Terrain Classification of Hyperspectral Remote Sensing Images.

In Computational intelligence and neuroscience

To overcome the difficulty of automating and intelligently classifying the ground features in remote-sensing hyperspectral images, machine learning methods are gradually introduced into the process of remote-sensing imaging. First, the PaviaU, Botswana, and Cuprite hyperspectral datasets are selected as research subjects in this study, and the objective is to process remote-sensing hyperspectral images via machine learning to realize the automatic and intelligent classification of features. Then, the basic principles of the support vector machine (SVM) and extreme learning machine (ELM) classification algorithms are introduced, and they are applied to the datasets. Next, by adjusting the parameter estimates using a restricted Boltzmann machine (RBM), a new terrain classification model of hyperspectral images that is based on a deep belief network (DBN) is constructed. Next, the SVM, ELM, and DBN classification algorithms for hyperspectral image terrain classification are analysed and compared in terms of accuracy and consistency. The results demonstrate that the average detection accuracies of ELM on the three datasets are 89.54%, 96.14%, and 96.28%, and the Kappa coefficient values are 0.832, 0.963, and 0.924; the average detection accuracies of SVM are 88.90%, 92.11%, and 91.68%, and the Kappa coefficient values are 0.768, 0.913, and 0.944; the average detection accuracies of the DBN classification model are 92.36%, 97.31%, and 98.84%, and the Kappa coefficient values are 0.883, 0.944, and 0.972. The results also demonstrate that the classification accuracy of the DBN algorithm exceeds those of the previous two methods because it fully utilizes the spatial and spectral information of hyperspectral remote-sensing images. In summary, the DBN algorithm that is proposed in this study has high application value in object classification for remote-sensing hyperspectral images.

Li Yanyi, Wang Jian, Gao Tong, Sun Qiwen, Zhang Liguo, Tang Mingxiu


General General

Modeling and forecasting the spread tendency of the COVID-19 in China.

In Advances in difference equations

To forecast the spread tendency of the COVID-19 in China and provide effective strategies to prevent the disease, an improved SEIR model was established. The parameters of our model were estimated based on collected data that were issued by the National Health Commission of China (NHCC) from January 10 to March 3. The model was used to forecast the spread tendency of the disease. The key factors influencing the epidemic were explored through modulation of the parameters, including the removal rate, the average number of the infected contacting the susceptible per day and the average number of the exposed contacting the susceptible per day. The correlation of the infected is 99.9% between established model data in this study and issued data by NHCC from January 10 to February 15. The correlation of the removed, the death and the cured are 99.8%, 99.8% and 99.6%, respectively. The average forecasting error rates of the infected, the removed, the death and the cured are 0.78%, 0.75%, 0.35% and 0.83%, respectively, from February 16 to March 3. The peak time of the epidemic forecast by our established model coincided with the issued data by NHCC. Therefore, our study established a mathematical model with high accuracy. The aforementioned parameters significantly affected the trend of the epidemic, suggesting that the exposed and the infected population should be strictly isolated. If the removal rate increases to 0.12, the epidemic will come to an end on May 25. In conclusion, the proposed mathematical model accurately forecast the spread tendency of COVID-19 in China and the model can be applied for other countries with appropriate modifications.

Sun Deshun, Duan Li, Xiong Jianyi, Wang Daping


COVID-19, Control strategy, Forecasting, Mathematical modeling, Parameter estimation