Interactive case session using PredictND tool AEC 2017, Berlin, Germany Hanneke Rhodius- Meester Clinical geriatrician VUmc Alzheimer Center, Amsterdam h.rhodius@vumc.nl
Disclosures The Alzheimer Center VUmc has received unrestricted funding from: Aegon, Alzheimer Nederland, VUmc Fonds, Heineken Nederland, Kroonenberg NV, Rabobank Amsterdam, Genootschap tot Steun Alzheimercentrum, Alzheimer Rally, and many others Contract research partners: GE Healthcare, Piramal Imaging, Merck, Nutricia, Probiodrug, EIP Pharma, Roche AG Espria, Combinostics, and Boehringer Ingelheim Grant funding: NWO, ZonMW, CVON, EU-FP7, EU-JPND, EU-IMI, ISAO, Alzheimer Nederland, Dioraphte, Gieskes Strijbis Dr Rhodius receives no personal compensation from any of the above or others except the Vumc; she is appointed on a grant from the European Seventh Framework Program project PredictND under grant agreement 611005.
Dementia diagnosis. Background
Background Dementia diagnosis.complex! Challenge as a clinician How to take into account all information? How to weigh and combine data? And. Literature often about controls vs Alzheimer Clinical practice is about differential diagnosis
Background To clinical practice. 73 year old male: History: transient ischemic attack Memory and word finding problems since 3 years, plus subtle hallucinations Mild parkinsonism, MMSE 28/30 Impairment in memory, naming and executive functions, slow MRI: mild atrophy and mild vascular damage No CSF obtained???
Background To clinical practice. 73 year old male: History: transient ischemic attack Memory and word finding problems since 3 years, plus subtle hallucinations Mild parkinsonism, MMSE 28/30 Impairment in memory, naming and executive functions, slow MRI: mild atrophy and mild vascular damage No CSF obtained Lewy body dementia? Alzheimer s disease?
Aim Could a computer tool help? Combining and visualizing data providing overview Dealing with conflicting data like clinical practice Comparing different dementia types with each other about differential diagnosis clinical practice is Mattila 2012, Antilla 2013, Rhodius- Meester 2016
Methods PredictND project computer tool Clinical descision support system: Combines and visualizes data to support clinician Entrance patient data Analyses tool How? Entrance baseline data of 1 patient: - demographics - CSF - cognitive tests - raw MRI images Outcome DSI and fingerprint
Methods Automatic MRI analyses Clinical practice: Majority of information hidden in MRI How to extract? With tool: Provides simple overview Highlights important areas Quantitative features, f.e. - hippocampal volumes Entrance patient data Analyses tool Outcome DSI and fingerprint Koikkalainen, NeuroImage Clin 2016
Methods Disease state index Compares patient baseline data with dataset Which diagnosis fits the data best? Entrance patient data Analyses tool Results in disease state index (DSI): Value between 0 and 1 cut-off 0.5 xx = 0= resembles f.e. controls xx = 1= resembles f.e. vascular dementia Outcome DSI and fingerprint
Methods Fingerprint Combines and visualises all data of 1 patient Size of box: xxxx = large contribution to diagnosis = small contribution Colour of box: xx = towards 0 (f.e. controls) xx = towards 1 (f.e. AD) Entrance patient data Analyses tool Outcome DSI and fingerprint
Translation to daily practice
Results Back to clinical practice. 73 year old male: History: transient ischemic attack Memory and word finding problems since 3 years, plus subtle hallucinations Mild parkinsonism, MMSE 28/30 Impairment in memory, naming and executive functions, slow MRI: MTA 1/1, GCA 1, Fazekas 1 all mild No CSF obtained Lewy body dementia? Alzheimer s disease?
Results 73 year old male tool Ingevoerde gegevens Ruwe MRI beelden Resultaten tool
Results 73 year old male Mixed pathology. Amyloïd-PET: positive DLB, with comorbid Alzheimer s pathology
Results 58 years old female Since 1 year complaints of forgetfulness, inappropriate behavior and urgency to speak. Interference in daily living No medical history Neuropsychology: MMSE 26/30, impairments in memory and language, difficult to stop MRI brain: atrophy frontal and temporal regions CSF: normal amyloid, tau and p-tau???
Results 58 years old female Since 1 year complaints of forgetfulness, inappropriate behavior and urgency to speak. Interference in daily living No medical history Neuropsychology: MMSE 26/30, impairments in memory and language, difficult to stop MRI brain: atrophy frontal and temporal regions CSF: normal amyloid, tau and p-tau FTLD, Alzheimer s disease?
58 year old female tool Results
Results 58 years old female Results tool DLB DSI 0.74 Behavioural variant of FTD AD DSI 0.59 xx = DLB, xx = AD
Results 71 years old female Memory complaints since 4 years, desorientation and apraxia, interference in cooking, shopping and driving. History of hypertension and diabetes medication Neuropsychology: MMSE 20/30, impairment in memory, executive functions, naming and visual functions. MRI brain: moderate hippocampal and parietal atrophy No CSF obtained???
Results 71 years old female Memory complaints since 4 years, desorientation and apraxia, interference in cooking, shopping and driving. History of hypertension and diabetes medication Neuropsychology: MMSE 20/30, impairment in memory, executive functions, naming and visual functions. MRI brain: moderate hippocampal and parietal atrophy No CSF obtained Alzheimer s dementia? Vascular dementia?
71 year old female tool Results
Results 71 years old female Results tool DLB DSI 0.74 Alzheimer s disease AD DSI 0.59 xx = DLB, xx = AD
Conclusion Assistance by computer Differential diagnosis of dementia is complex, computer tool can help by: Combining and visualising data provides overview Adding fully automated MRI analyses Multiclass tool daily practice is about differential diagnosis and multiple pathologies Yet, difficult cases will remain difficult: Tool depends on quality of data shows what you enter Can never replace the clinician only supports
v h.rhodius@vumc.nl