Application of fractal dimension for quantifying noise texture in computed tomography images

Khobragade, P.; Fan, Jiahua; Rupcich, Franco; Crotty, Dominic J.; Schmidt, Taly Gilat

doi:10.1002/mp.13040

Cited by 10 publications

(4 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This finding can be explained by the fact that increasing the ROI results increases the number of scale factors D , i.e., more data points for FD estimation. Also, the number of boxes per scale is increased, thus resulting in a more stable computation of FD [ 19 ].…”

Section: Discussionmentioning

confidence: 99%

Fractal dimension in CBCT images as predictor for MRONJ: a retrospective cohort study

Bachtler

Walter²,

Schulze

2020

Clin Oral Invest

View full text Add to dashboard Cite

Objective To estimate boxcounting fractal dimension in a standardized approach in CBCT images of the mandible and maxilla in a group of patients with MRONJ in comparison to a healthy control group. Material and methods From our records, a study group consisting of 80 maxillary and mandibular CBCTs in 77 cases of patients having MRONJ was collected. The control group consisted of 84 mandibular and maxillary CBCTs in a total of 78 patients. Using the boxcounting method, fractal dimension (FD) was estimated in a standardized fashion either cranially to the apex of the canine (maxilla) or beneath the tooth apices of the lower molars in the bone area above the mandibular canal (mandible). Intraobserver reproducibility of the FD-measurements was assessed by 6 repeated measurements in 10 individuals. FD values were correlated to age and sex of the individuals as well as to region of interest (ROI) sizes. Results FD in the study group (1.684 ± 0.051) was roughly 3.5 % lower than in the control group (1.745 ± 0.026, p < 0.0001). Sex and age had a significant (p < 0.001) influence on FD values in the study group, yet not in the control group. FD values increased with age (study group, spearman-rho: 0.2895, p < 0.05) and also ROI size (both groups, p < 0.0001). Reproducibility was good (intra-class correlation coefficient (ICC): 0.87). Conclusions Fractal dimension as assessed by boxcounting seems to be a good descriptor for MRONJ in jaw bones. Influence of age and sex on the outcome values needs to be further investigated in future studies. Clinical relevance CBCTs could be assessed with respect to FD to obtain an overview of the disease status of MRONJ patients.

show abstract

Section: Discussionmentioning

confidence: 99%

Fractal dimension in CBCT images as predictor for MRONJ: a retrospective cohort study

Bachtler

Walter²,

Schulze

2020

Clin Oral Invest

View full text Add to dashboard Cite

show abstract

“…The natural images are composed by self-similar elements, then, all natural image can be considered as a fractal set. Investigators have used the FD for many applications: quantification of noises (3), segmentation of images (4,5), analysis of textures (6), recognition of patterns (7), biometrics (8), ecology (9), cancer detection (10) and tomography (11).…”

Section: Introductionmentioning

confidence: 99%

Fractal Analysis in Medical Images

Pérez Labardi

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The investigation procedure we adopted was found efficient enough to accurately detect and quantify the differences in CNNs behavior both for the attained results and for the alterations introduced in the processed images. Its sensitivity has also proved adequate to properly quantify the differences in noise shape associated to the reconstruction methods: Filtered Back Projections (FBP) and Iterative Reconstruction (IR) [62][63][64].…”

Section: Introductionmentioning

confidence: 99%

Addressing signal alterations induced in CT images by deep learning processing: A preliminary phantom study

et al. 2021

View full text Add to dashboard Cite

We investigate, by an extensive quality evaluation approach, performances and potential side effects introduced in Computed Tomography (CT) images by Deep Learning (DL) processing. Method: We selected two relevant processing steps, denoise and segmentation, implemented by two Convolutional Neural Networks (CNNs) models based on autoencoder architecture (encoder-decoder and UNet) and trained for the two tasks. In order to limit the number of uncontrolled variables, we designed a phantom containing cylindrical inserts of different sizes, filled with iodinated contrast media. A large CT image dataset was collected at different acquisition settings and two reconstruction algorithms. We characterized the CNNs behavior using metrics from the signal detection theory, radiological and conventional image quality parameters, and finally unconventional radiomic features analysis. Results: The UNet, due to the deeper architecture complexity, outperformed the shallower encoder-decoder in terms of conventional quality parameters and preserved spatial resolution. We also studied how the CNNs modify the noise texture by using radiomic analysis, identifying sensitive and insensitive features to the denoise processing. Conclusions: The proposed evaluation approach proved effective to accurately analyze and quantify the differences in CNNs behavior, in particular with regard to the alterations introduced in the processed images. Our results suggest that even a deeper and more complex network, which achieves good performances, is not necessarily a better network because it can modify texture features in an unwanted way.

show abstract

Application of fractal dimension for quantifying noise texture in computed tomography images

Cited by 10 publications

References 28 publications

Fractal dimension in CBCT images as predictor for MRONJ: a retrospective cohort study

Fractal dimension in CBCT images as predictor for MRONJ: a retrospective cohort study

Fractal Analysis in Medical Images

Addressing signal alterations induced in CT images by deep learning processing: A preliminary phantom study

Contact Info

Product

Resources

About