Identification of the reporter gene combination that shows high contrast for cellular level MRI

by Naoya Hayashi, Junichi Hata, Tetsu Yoshida, Daisuke Yoshimaru, Yawara Haga, Hinako Oshiro, Ayano Oku, Noriyuki Kishi, Takako Shirakawa, Hideyuki Okano

Currently, we can label the certain cells by transducing specific genes, called reporter genes, and distinguish them from other cells. For example, fluorescent protein such as green fluorescence protein (GFP) is commonly used for cell labeling. However, fluorescent protein is difficult to observe in living animals. We can observe the reporter signals of the luciferin-luciferase system from the outside of living animals using in vivo imaging systems, although the resolution of this system is low. Therefore, in this study, we examined the reporter genes, which allowed the MRI-mediated observation of labeled cells in living animals. As a preliminary stage of animal study, we transduced some groups of plasmids that coded the protein that could take and store metal ions to the cell culture, added metal ions solutions, and measured their T1 or T2 relaxation values. Finally, we specified the best reporter gene combination for MRI, which was the combination of transferrin receptor, DMT1, and Ferritin-M6A for T1WI, and Ferritin-M6A for T2WI.

Inter- and intra-rater reproducibility of quantitative T1 measurement using semiautomatic region of interest placement in myometrium

by Sadahiro Nakagawa, Takahiro Uno, Shunta Ishitoya, Eriko Takabayashi, Akiko Oya, Wakako Kubota, Atsutaka Okizaki

Purpose

This study aimed to investigate the inter- and intraobserver reproducibility of quantitative T1 (qT1) measurements using manual and semiautomatic region of interest (ROI) placements. We hypothesized the usefulness of the semiautomatic method, which utilizes a three-dimensional (3D) anatomical relationship between the myometrium and other tissues, for minimizing ROI placement variation, thereby improving qT1 reproducibility compared to the manual approach. The semiautomatic approach, which considered anatomical relationships, was expected to enhance reproducibility by reducing ROI placement variabilities.

Materials and methods

This study recruited 23 healthy female volunteers. Data with variable flip angle (VFA) and inversion recovery were acquired using 3D-spoiled gradient echo and spin echo sequences, respectively. T1 maps were generated with VFA. Manual and semiautomatic ROI placements were independently conducted. Mean qT1 values were calculated from the T1 maps using the corresponding pixel values of the myometrial ROI. Inter- and intraobserver reproducibility of qT1 values was investigated. The inter- and intraobserver reproducibility of qT1 values was evaluated by calculating the coefficient of variation (CoV). Further, reproducibility was evaluated with inter- and intraobserver errors and intraclass correlation coefficients (ICCs). Bland–Altman analysis was utilized to compare the results, estimate bias, and determine the limits of agreement.

Results

The mean inter- and intraobserver CoV of the qT1 values for semiautomatic ROI placement was significantly lower than those for manual ROI placement (p p p p Conclusion

Semiautomatic ROI placement demonstrated high reproducibility of qT1 measurements compared with manual methods. Semiautomatic ROI placement may be useful for evaluating uterine qT1 with high reproducibility.

A comparative analysis of converters of tabular data into image for the classification of Arboviruses using Convolutional Neural Networks

by Leonides Medeiros Neto, Sebastião Rogerio da Silva Neto, Patricia Takako Endo

Tabular data is commonly used in business and literature and can be analyzed using tree-based Machine Learning (ML) algorithms to extract meaningful information. Deep Learning (DL) excels in data such as image, sound, and text, but it is less frequently utilized with tabular data. However, it is possible to use tools to convert tabular data into images for use with Convolutional Neural Networks (CNNs) which are powerful DL models for image classification. The goal of this work is to compare the performance of converters for tabular data into images, select the best one, optimize a CNN using random search, and compare it with an optimized ML algorithm, the XGBoost. Results show that even a basic CNN, with only 1 convolutional layer, can reach comparable metrics to the XGBoost, which was trained on the original tabular data and optimized with grid search and feature selection. However, further optimization of the CNN with random search did not significantly improve its performance.