Visualization methods for analysis of 3D multi-scale medical data

 Ri­car­do Millán, Leib­niz Uni­ver­si­tät Han­no­ver

This the­sis was de­ve­l­o­ped wi­t­hin the frame­work of the EU Mul­tiS­cal­e­Hu­man Pro­ject, whose cen­tral aim lies on im­pro­ving the cur­rent un­der­stan­ding of phy­sio­lo­gi­cal human ar­ti­cu­la­ti­on with the ap­p­li­ca­ti­on of sup­porting the dia­gno­sis and tre­at­ment of mus­cu­los­kel­etal di­sea­ses. Towards this goal, this work cont­ri­bu­tes se­veral me­thods in order to im­pro­ve the vi­sua­liza­t­i­on and pro­ces­sing of bio­me­di­cal data. It es­pe­cial­ly ad­dres­ses sci­en­tists whose work­flow in­vol­ves po­ten­ti­al­ly mas­si­ve amount of bio­me­di­cal in­for­ma­ti­on.

In order to over­co­me the li­mi­ta­ti­ons in the cur­rent state of the art, this work cont­ri­bu­tes con­cepts for mul­ti-sca­le vi­sua­liza­t­i­on, par­ti­cu­lar­ly pro­vi­ding a 3D mul­ti­laye­red en­vi­ron­ment, in which glo­bal and de­tai­led in­for­ma­ti­on from mul­ti-sca­le bio­me­di­cal data can be pre­sen­ted and ex­plo­red, a com­bi­na­ti­on of sci­en­ti­fic and in­for­ma­ti­on vi­sua­liza­t­i­on, which aims to over­co­me the li­mi­ta­ti­on of using one group of vi­sua­liza­t­i­on me­thods to re­pre­sent bio­me­di­cal in­for­ma­ti­on with dif­fe­rent le­vels of ab­strac­tion, the in­tro­duc­tion of se­man­tic in­for­ma­ti­on from the mul­ti-sca­le data for the crea­ti­on, cust­o­miza­t­i­on and en­rich­ment of the mul­ti-sca­le vi­sua­liza­t­i­on, and a set of en­ri­ched in­tra-sca­le vi­sua­liza­t­i­ons, en­han­cing the pre­sen­ta­ti­on and ana­ly­sis of data wi­t­hin the con­side­red scale in order to sup­port the cli­ni­cal work­flow. In this con­text, this the­sis pres­ents con­cepts for the ana­ly­sis of ki­ne­ma­ti­cal data of human joint ar­ti­cu­la­ti­on at the be­ha­vioral scale.

The mul­ti-sca­le vi­sua­liza­t­i­on frame­work has the po­ten­ti­al to am­pli­fy the un­der­stan­ding of bio­me­di­cal pro­ces­ses in the human body. This has been con­fir­med in an ap­p­li­ca­ti­on sce­na­rio fo­cu­sing on the in­ves­ti­ga­ti­on of knee joint ki­ne­ma­tics based on ex­pe­ri­men­tal data collec­ted in col­la­bo­ra­ti­on with MHH-LBB, the lat­ter being a pro­ject part­ner wi­t­hin the Mul­tiS­cal­e­Hu­man pro­ject. It was de­mons­tra­ted, that the de­ve­l­o­ped pro­ces­sing me­thods re­ly­ing on Lie group theo­ry and the cor­re­spon­ding vi­sua­liza­t­i­on tech­ni­ques allow for an in­tui­ti­ve in­ter­pre­ta­ti­on of he­li­cal axes and their geo­metri­cal re­la­ti­on with re­spect to the knee joint ana­to­my, an ef­fi­ci­ent dif­fe­ren­tia­ti­on among se­veral cases and sta­tes of knee joint, and a fle­xi­ble ex­plo­ra­ti­on of the dis­cus­sed data sets and of re­la­ted data, the­r­e­by sup­porting the cli­ni­cal work­flow du­ring the ana­ly­sis of the knee joint ar­ti­cu­la­ti­on.