www.bayesia.com does not fully support your browser (Internet Explorer 6).
We suggest upgrading to IE 7 or downloading Firefox for a more enjoyable web experience.

Breakthroughs in radiology thanks to BayesiaLab

November, 15 2005 / press release

BayesiaLab, a data analysis software based on Bayesian networks, has allowed crucial medical breakthroughs and developments in interventional radiology. The highlighting of unthought-of correlations is at the basis of these advances. Bayesia Ltd. ’s idiosyncratic method of data analysis could easily be applied to other research fields, in radiology as well as in medicine in general.

Marc Legeais, assistant senior registrar at the radiology department of Tours University Hospital (Pr Herbreteau), France, has appealed to Bayesia’s experts to use their data analysis software in order to finalize his Ph.D thesis. Indeed, BayesiaLab’s unique features allowed his research to go much further than with any other data analysis tool. New reflection perspectives have emerged and been explored.
« BayesiaLab’s unsupervised analysis of our database has brought to light correlations we would never have thought of, or even researched ! That is the reason why this method is so innovatory and profoundly interesting.»

How did you get in touch with Bayesia ?

Marc Legeais : « At first, some members of the ESIEA (a school of engineering that provides education in the fields of Information Technology, Electronics and Automatic Systems, in which Lionel Jouffe used to teach and conduct his research) alluded to the company, and then I logged on to their website. I started to foresee the potential applications of Bayesia thanks to the extremely didactic nature of the website, with the possibility to download a great many Flash animations. »

Regarding your PhD thesis, what were you expecting ?

Marc Legeais : « I had to find a statistics tool that would shed a relevant light on the analysis of an huge databank. »

What did BayesiaLab bring ?

ML : « BayesiaLab allowed me to go much further than I initially planned. I could do a series of original applications: an intelligent imputation of missing data, search for correlations buried in my database, characterisation of identified variables (such as for instance the result of a therapeutic act or the number of recurrences after treatment).

On top of these applications, BayesiaLab offers us a real time simulation tool we can use during consultations to inform women about the “chances” of statistic evolution with respect to the initial variables we have (the number of fibromas, their size, the patient’s age, the number of children she has…).

In the long run, our use of BayesiaLab is to allow us to modify our treatment protocols so as to favour specific, better adapted treatments for our patients and respect their personal features (for example : the number of flasks of embolisation particles to be used in relation to the size of the uterus,the number of fibromas and the size of the dominant fibroma).»

In the light of the work that Bayesia’s experts have done for your thesis, are there any extensions or new applications possible ?

ML : « Indeed, industries (manufacturing medical equipment) do need a software that they may distribute for their own benefit indexed to our databases and that would allow the doctors who buy their equipement to get a statistic predictive tool.

The use of BayesiaLab lets us cast a new and critical glance on our exercise in interventional radiology and identify the influential variables. »

Marc Legeais’s thesis defense has attracted a lot of attention, especially thanks to the achievements of Bayesia’s software. Doctor Legeais has also presented his work at the 2005 French Radiology Congress. Moreover, The « French Review of Radiology » (the first reference in terms of scientific information source for French radiologists) would like to publish an article about this new method of medical data analysis. « Indeed, Lionel Jouffe and I want to highlight this data analysis method that can also be applied to other radiology or medicine fields ».

Further Information

The title and precise theme of Marc Legeais’s thesis are : « Uterine Embolised Leiomyoma : MRI aspects, Anatomopathological Correlations, Proposition for a Radiohistological Classification, Integration in a Bayesian Network. »

Uterus fibromas (or leiomyomas) are benign tumors that develop in the uterus of women, most of them over 40. In France, less than one woman out of three is concerned. These leiomyomas can cause bleeding, pain, compression of the neighbouring organs (bladder, rectum) and finally infertility. To date, women had to go through surgery, and their uterus was fully removed (this year, 75000 hysterectomies have been carried out in France).

Embolisation (i.e. the obstruction of uterine arteries by the endovascular tract) makes it possible to avoid such a mutilating surgery and therefore to respect women’s integrity and help relieve their symptoms.

Tours University Hospital is one of the leaders in embolisation of uterine leiomyoma, in France and worldwide.