Try our AI tools for analysis of PET/CT and CT studies

We invite researchers to apply and evaluate our AI tools on their own PET/CT or CT studies. We are also happy to collaborate and develop customized AI tools for other tasks than those presented below.

Request access to our platform by mail to contact@recomia.org.

This video shows how to upload and apply AI tools to your CT or PET/CT studies.

NOTE: The AI tools are available for research purposes only, not for clinical use.

Body Composition CT
Our Body Composition is based on the AI tool “Organ Finder version 5”, which outputs 3D-segmentations of muscle, subcutaneous adipose tissue, and visceral adipose tissue in CT studies in addition to many other organs. Organ Finder is trained on more than 1,250 CT studies with manual segmentations of more than 10,000 organs and tissues. The whole CT is segmented by the Organ Finder. If only a defined part of the body should be studied, for example the thorax, abdomen, or torso, a separate quantification tool is applied afterwards.
The method has been described in a paper by Borrelli (2021) and in a recent paper Ying (2021) showed that muscle volume measured with this AI tool was an independent biomarker for survival in patients who have undergone radical cystectomy. Most recently Ying (2024) showed the value of 3D segmentations compared to analysis in a single CT slice.

Prostate Cancer
We have developed several versions of our AI-based tools for analysis of PSMA-PET/CT studies, each time improving training methods and increasing training databases. Currently the training database includes more than 1,000 PSMA-PET/CT studies based on different PSMA tracers. A previous version has been described by Trägårdh (2022). Current work has the aim to train an AI tool which can handle studies of different PSMA tracers.

Organ Activity in PET/CT – a dosimetry tool
PET tracer uptake in different organs can be measured based on an organ segmentation of CT produced by the AI tool “Organ Finder version 5”. This AI tool segments 20-30 different soft tissue organs and 20-30 different bones in CT. The tool is trained on more than 1,250 CT studies with more than 10,000 manual annotations. The CT studies are selected from public databases with origin from 10 different hospitals, 4 different scanner types (GE, Philips, Siemens and Toshiba), varying fields of view (abdomen, chest, torso, …), and varying intravenous/oral contrast – with and without in different phases. A previous version of Organ Finder was presented by Edenbrandt (2022).

Lymphoma
Tumour lesion glycolysis (TLG) and metabolic tumour volume (MTV) have shown to be associated with progression-free and overall survival in Hodgkin ́s lymphoma patients. An automated and objective quantification of TLG and MTV can be obtained using our AI tool “Lymphoma FDG-PET/CT version 2”. Lymph nodes with increased FDG uptake, focal uptake in spleen, bone marrow and other extranodal sites are detected and segmented. A comparison between AI-based and manual MTV measures was recently presented by Sadik (2024).

Multiple myeloma
A volumetric quantification of bone marrow metabolism can be achieved using a combination of bone segmentation in CT, analysis of PET activity in the corresponding PET voxels, and application of different SUV thresholds. The bone marrow metabolism has shown to predict survival in multiple myeloma patients. The results have been presented by Sachpekidis (2023 and 2024).

Atherosclerosis
Our AI tool “Heart and Aortic Wall version 1” and “Carotid segmentation version 1” can be used to quantify F-sodium fluoride (NaF) activity in the heart, aortic wall, and carotid as signs of microcalcifications. The AI tools segment the aortic arch, the abdominal, and the thoracic parts of the aorta as well as the heart and carotids in the CT images. The spine is also segmented in CT allowing high PET activity originating from the spine to be excluded from the aortic wall segmentations. A comparison between AI-based and manual carotid segmentations was recently presented by Piri (2024).

Lung Cancer
Our AI tool “Lung Cancer FDG-PET/CT v1.2” detects lung tumors, thoracic lymph nodes, and bone metastases in FDG-PET/CT studies. Note that lymph nodes outside thorax are not detected in this version. The AI tool includes an organ segmentation module and is trained on 320 FDG-PET/CT studies with manual annotations from patients with suspected lung cancer. An earlier version of this tool was presented in a paper by Borrelli (2022).