Application of neural networks for improved diagnosis of pulmonary artery embolism in ventilation/perfusion SPECT/CT and possible omission of the ventilation component

• Applicant: Dr. David Kersting
• Subject Area: Radiology
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 519060267

Pulmonary embolism is a common and potentially life-threatening disease whose diagnosis and follow-up assessment are of high importance. The clinical presentation is highly variable, ranging from asymptomatic patients to sudden death. To prevent a severe course and complications, a rapid diagnosis and a quick start of therapy are essential. Despite optimal care, complications such as recurrent disease and chronic thromboembolic pulmonary hypertension are possible. Typical algorithms recommend ventilation/perfusion-SPECT (V/P-SPECT) and CT pulmonary angiogram (CTPA) as imaging diagnostics. The advantages of V/P-SPECT are a higher applicability due to fewer contraindications, a higher sensitivity with comparable specificity, lower radiation exposure and the possibility of quantifying the affected lung parenchyma, which is important for follow-up examinations. However, the availability of the examination is lower. In particular, the ventilation-SPECT requires expensive radionuclide generators, which limits the availability of examinations. In addition, ventilation-SPECT is also critical in the ongoing COVID-19 pandemic, as its performance exposes medical staff to a high risk of infection when examining infected patients. An alternative is to perform a pure perfusion-SPECT without ventilation-SPECT. Thus, perfusion defects can be detected, and the ventilated lung parenchyma can be assessed on a CT scan acquired in hybrid imaging technique in the same examination (SPECT/CT). Hence, a high sensitivity comparable to V/P-SPECT can be reached, but the specificity is significantly limited. This can lead to a high rate of false-positive findings as well as an overdiagnosis of pulmonary embolism. Thus, there is still an unmet clinical need to improve V/P-SPECT and pure perfusion-SPECT/CT examinations. The high number of recent publications shows growing scientific interest in artificial intelligence (AI)-based methods for the analysis of nuclear medicine images by neural networks to enable automatic image evaluation and improved diagnostics. To date, there is only insufficient data on such applications to V/P-SPECT data for the diagnosis of pulmonary embolism. Therefore, the hypothesis of this project is that sophisticated evaluation of V/PSPECT/ CT using AI-based methods has the potential to significantly increase diagnostic quality and to increase the specificity of pure perfusion-SPECT/CT making ventilation-SPECT unnecessary for the diagnosis of pulmonary embolism.

• DFG Programme: Research Grants
• Co-Investigators: Professor Dr. Christoph Rischpler; Dr. Robert Seifert