http://www.isnvdconference.org/program/ ... -book.html
12 FEBRUARY 7 – 9, 2014 | ISNVD 2014
Identifying CCSVI with Cervical Plethysmography. TELEPASS Study and Kosmomed Project Presentation
Paolo Zamboni, M.D., Simon Shepherd, Clive Beggs, Francesco Speziale, Robert Zivadinov, Pierfrancesco Veroux, Stefano Bastianello, Michael D. Dake, Fabrizio Salvi, Mariano Bizzarri.
Background: A significant heterogeneity of prevalence studies on chronic cerebrospinal venous insufficiency (CCSVI) by the means of Doppler ultrasound (DUS) generates a significant scientific controversy.
Recently, it has been also developed a cervical plethysmography (CP) in order to objectively measure the cerebral venous return. Given that the cervical plethysmography technique produces clear quantifiable results this suggests that the new technique has great potential as an inexpensive diagnostic tool. Kosmomed is a telemedicine project supported by the European Space Agency, which approved the Telepass study acronym of Telemedicine Plethysmography Cerebral Venous Return Assessment. This study is ongoing and represents an opportunity for experimenting a CCSVI screening program.
Objectives: To investigate sensitivity and specificity of CP performed by a technician and interpreted in remote fashion. Results of cervical plethysmography will be compared in patients and controls to the results of a multi modal protocol recently proposed (Premise trial).
Methods: Thirty (30) healthy controls matched for age and gender with 70 MS patients, and 50 patients affected by other neurological diseases (OND). They will be blindly screened by the means of CP. Subsequently, subjects of all 3 groups will be blindly evaluated with multimodal protocol including: DUS, MRV, Catheter venography (CV) and IVUS. All image modalities will be analyzed in a blinded manner by more than one viewer, upon which consensus will be reached. The sensitivity and specificity will be calculated using contingency tables denoting the presence or absence of vein-specific abnormality findings between all imaging modalities used individually as the gold standard.
How to improve diagnostic accuracy: Currently, CP sensitivity and specificity are both around 70%. We aim to improve this very considerably using SVM machine-learning techniques. In simple terms, an SVM seeks to translate the low dimensional vector space where the data resides (called the input space) to a higher dimensional vector space (called the feature space) in order to make clustering (and hence classification) easier and more powerful. At all stages of the project, the diagnostic results will be validated by the “gold standard” multimodality techniques.
This will permit to know the diagnostic accuracy of CP in telemedicine modality as a CCSVI screening device and to obtain final validation of CP as screening system.
I very much like the multimodal protocol. Patients will be screened first with plethysmography, and then Doppler ultrasound, MRV, catheter venography and IVUS.
I don't think it's good, though, that plethysmography sensitivity and specificity are both at 70%. That's lower than I would have expected. In a clinical setting, you don't want to screen out people as not having CCSVI when they actually do have CCSVI. And in a research setting also, it is important for it to be a highly effective diagnosis tool. SVM machine-learning techniques might improve this but even the "in simple terms" explanation is rather complex for me.
By comparing the plethysmography results to the multimodal results, it does give me confidence in the results.