3 results
2555 Predictive cytological topography (PiCT): A radiopathomics approach to mapping prostate cancer
- Sean D. McGarry, Sarah L. Hurrell, Kenneth Ickzkowski, Anjishnu Banerjee, Kenneth Jacobsohn, William Hall, Mark Hohenwalter, Peter LaViolette, Amy Kaczmarowski, Tucker Keuter, Marja Nevalainen, William See
-
- Journal:
- Journal of Clinical and Translational Science / Volume 2 / Issue S1 / June 2018
- Published online by Cambridge University Press:
- 21 November 2018, pp. 23-24
-
- Article
-
- You have access Access
- Open access
- Export citation
-
OBJECTIVES/SPECIFIC AIMS: The objective of this study is to use machine Learning techniques to generate maps of epithelium and lumen density in MRI space. METHODS/STUDY POPULATION: Methods: We prospectively recruited 39 patients undergoing prostatectomy for this institutional review board (IRB) approved study. Patients underwent MP-MRI before prostatectomy on a 3T field strength MRI scanner (General Electric, Waukesha, WI, USA) using an endorectal coil. MP-MRI included field-of-view optimized and constrained undistorted single shot (FOCUS) diffusion weighted imaging with 10 b-values (b=0, 10, 25, 50, 80, 100, 200, 500, 1000, and 2000), dynamic contrast enhanced imaging, and T2-weighted imaging. T2 weighted images were intensity normalized and apparent diffusion coefficient maps were calculated. The dynamic contrast enhanced data was used to calculate the percent change in signal intensity before and after contrast injection. All images were aligned to the T2 weighted image. Robotic prostatectomy was performed 2 weeks after image acquisition. Prostate samples were sliced using a 3D printed slicing jig matching the slice profile of the T2 weighted image. Whole mount samples at 10 μm thickness were taken, hematoxylin and eosin stained, digitized, and annotated by a board certified pathologist. A total of 210 slides were included in this study. Lumen and epithelium were automatically segmented using a custom algorithm written in MATLAB. The algorithm was validated by comparing manual to automatic segmentation on 18 samples. Slides were aligned with the T2 weighted image using a nonlinear control point warping technique. Lumen and epithelium density and the expert annotation were subsequently transformed into MRI space. Co-registration was validated by applying a known warp to tumor masks noted by the pathologist and control point warping the whole mount slide to match the transform. Overlap was measured using a DICE coefficient. A learning curve was generated to determine the optimal number of patients to train the algorithm on. A PLS algorithm was trained on 150 random permutations of patients incrementing from 1 to 29 patients. Slides were stratified such that all slides from a single patient were in the same cohort. Three cohorts were generated, with tumor burden balanced across all cohort. A PLS algorithm was trained on 2 independent training sets (cohorts 1 and 2) and applied to cohort 3. The input vector consisted of MRI values and the target variable was lumen and epithelium density. The algorithm was trained lesion-wise. Trained PiCT models were applied to the test cohort voxel-wise to generate 2 new image contrasts. Mean lesion values were compared between high grade, low grade, and healthy tissue using an ANOVA. An ROC analysis was performed lesion-wise on the test set. RESULTS/ANTICIPATED RESULTS: Results: The segmentation accuracy validation revealed R=0.99 and R=0.72 (p<0.001) for lumen and epithelium, respectively. The co-registration accuracy revealed a 94.5% overlap. The learning curve stabilized at 10 patients with a root mean square error of 0.14, thus the size of the 2 independent training cohorts was set to 10, leaving 19 for the test cohort. DISCUSSION/SIGNIFICANCE OF IMPACT: We present a technique for combining radiology and pathology with machine learning for generating predictive cytological topography (PiCT) maps of cellularity and lumen density prostate. The voxel-wise approach to mapping cellular features generates 2 new interpretable image contrasts, which can potentially increase confidence in diagnosis or guide biopsy and radiation treatment.
Food photographs in portion size estimation among adolescent Mozambican girls
- Liisa Korkalo, Maijaliisa Erkkola, Lourdes Fidalgo, Jaakko Nevalainen, Marja Mutanen
-
- Journal:
- Public Health Nutrition / Volume 16 / Issue 9 / September 2013
- Published online by Cambridge University Press:
- 08 August 2012, pp. 1558-1564
-
- Article
-
- You have access Access
- HTML
- Export citation
-
Objective
To assess the validity of food photographs in portion size estimation among adolescent girls in Mozambique. The study was carried out in preparation for the larger ZANE study, which used the 24 h dietary recall method.
DesignLife-sized photographs of three portion sizes of two staple foods and three sauces were produced. Participants ate weighed portions of one staple food and one sauce. After the meal, they were asked to estimate the amount of food with the aid of the food photographs.
SettingZambezia Province, Mozambique.
SubjectsNinety-nine girls aged 13–18 years.
ResultsThe mean differences between estimated and actual portion sizes relative to the actual portion size ranged from −19 % to 8 % for different foods. The respective mean difference for all foods combined was −5 % (95 % CI −12, 2 %). Especially larger portions of the staple foods were often underestimated. For the staple foods, between 62 % and 64 % of the participants were classified into the same thirds of the distribution of estimated and actual food consumption and for sauces, the percentages ranged from 38 % to 63 %. Bland–Altman plots showed wide limits of agreement.
ConclusionsUsing life-sized food photographs among adolescent Mozambican girls resulted in a rather large variation in the accuracy of individuals’ estimates. The ability to rank individuals according to their consumption was, however, satisfactory for most foods. There seems to be a need to further develop and test food photographs used in different populations in Sub-Saharan Africa to improve the accuracy of portion size estimates.
Some similarities in dietary clusters of pre-school children and their mothers
- Marja-Leena Ovaskainen, Jaakko Nevalainen, Liisa Uusitalo, Jetta J. Tuokkola, Tuula Arkkola, Carina Kronberg-Kippilä, Riitta Veijola, Mikael Knip, Suvi M. Virtanen
-
- Journal:
- British Journal of Nutrition / Volume 102 / Issue 3 / 14 August 2009
- Published online by Cambridge University Press:
- 02 March 2009, pp. 443-452
- Print publication:
- 14 August 2009
-
- Article
-
- You have access Access
- HTML
- Export citation
-
The diet of pre-school children is determined by the parents and carers. The aim of the present study was to describe dietary clusters of pre-school children and their mothers in Finland, and analyse the similarity of dietary clusters within child–mother pairs. The present study comprised the mothers (n 4862) whose child was recruited in the Type 1 Diabetes Prediction and Prevention Nutrition Study and the children belonging to selected, cross-sectional age groups of 1 year (n 719), 3 years (n 708) and 6 years (n 841). The dietary data were collected from children by 3-d food records and from mothers by a FFQ validated for pregnant women. The food consumption data were analysed for patterns by hierarchical cluster analysis. Three main dietary clusters were identified in children: ‘healthy’ and ‘traditional’ in all three age groups, and ‘ready-to-eat baby foods’ in 1-year-olds and ‘fast foods, sweet’ in the older children. Six main clusters were identified among the mothers who completed a FFQ for their diet during pregnancy. Some familial dependence between dietary clusters of mother–child pairs was observed in 6-year-old children but not in younger children. Younger age and lower educational level of the mother were associated with the cluster ‘fast food, sweet’ only at the age of 3 years. The diets of pre-school children vary by age and only a slight similarity within dietary clusters of mother–child pairs was observed.