Article of the Month – September 2021
September 1, 2021
Article of the Month – October 2021
October 1, 2021France
OWKIN
Innovation in AI at OWKIN is enormous. One of their projects sought to detect Lymphoma lesions and to perform segmentation on whole-body FDG-PET/CT. Detection and segmentation have proven to be challenging tasks due to the diversity of involved nodes, organs, or physiological uptakes. They sought to investigate the performance of a three-dimensional (3D) convolutional neural network (CNN) to automatically segment total metabolic tumor volume (TMTV) in large datasets of patients with diffuse large B cell lymphoma (DLBCL).
Innovation in AI at OWKIN is enormous. One of their projects sought to detect Lymphoma lesions and to perform segmentation on whole-body FDG-PET/CT. Detection and segmentation have proven to be challenging tasks due to the diversity of involved nodes, organs, or physiological uptakes. They sought to investigate the performance of a three-dimensional (3D) convolutional neural network (CNN) to automatically segment total metabolic tumor volume (TMTV) in large datasets of patients with diffuse large B cell lymphoma (DLBCL).
Segmentation of lymphoma lesions has proven to aid in treatment plan design, improving outcomes. They used large datasets from de novo DLBCL patients to train the CNN and an independent cohort of similar patients for external validation purposes. Despite a slight underestimation of TMTV, it promises to be an excellent asset for automated detection and segmentation of lymphoma lesions. Furthermore, its fully automated and open-source hallmarks may ease its widespread use in routine practice and promote reproducibility of TMTV assessment in different types of lymphoma.(1)
Primaa
Primaa developed an AI-enabled software to help pathologists detect cancer at microscopic levels undetectable to the expert eye. These programs improve early detection, a vital step in cancer care, by assisting pathologists with safe and accurate diagnoses. Although the algorithm is not yet completely autonomous, they aim for pathology exams to be delivered faster and more effectively.(2)
Primaa developed an AI-enabled software to help pathologists detect cancer at microscopic levels undetectable to the expert eye. These programs improve early detection, a vital step in cancer care, by assisting pathologists with safe and accurate diagnoses. Although the algorithm is not yet completely autonomous, they aim for pathology exams to be delivered faster and more effectively.(2)
AQEMIA
AQEMIA, on the other hand, aids in drug discovery by implementing AI to rapidly identify more innovative molecules with better success chances in drug manufacturing. Their software predicts accurately and efficiently the affinity between a compound and a therapeutic target. The AI algorithm training consists in learning physics and molecular chemistry rather than raw data.(3)
AQEMIA, on the other hand, aids in drug discovery by implementing AI to rapidly identify more innovative molecules with better success chances in drug manufacturing. Their software predicts accurately and efficiently the affinity between a compound and a therapeutic target. The AI algorithm training consists in learning physics and molecular chemistry rather than raw data.(3)
Gleamer
Gleamer, an AI-powered startup company in France, forges AI and radiology with its first product called BoneView. The application is used by more than 750 users, including radiologists and emergency physicians, in more than 50 hospitals with an output of 99.7% Negative Predictive Value for abnormal bone findings.(4)
Gleamer, an AI-powered startup company in France, forges AI and radiology with its first product called BoneView. The application is used by more than 750 users, including radiologists and emergency physicians, in more than 50 hospitals with an output of 99.7% Negative Predictive Value for abnormal bone findings.(4)
Mojo
Mojo, is revolutionizing in-vitro fertilization techniques and outcomes. Through 3 steps (Assess, Select, and Inject), their AI-assisted semen analysis, and their use of intelligent microscopes, they pick superior sperms to facilitate the IVF process. Their approach provides a higher success rate and, therefore, a more affordable approach to the devastating problem of infertility.(5)
Mojo, is revolutionizing in-vitro fertilization techniques and outcomes. Through 3 steps (Assess, Select, and Inject), their AI-assisted semen analysis, and their use of intelligent microscopes, they pick superior sperms to facilitate the IVF process. Their approach provides a higher success rate and, therefore, a more affordable approach to the devastating problem of infertility.(5)
References