Exploring An Easy Way for Imbalanced Data Sets in Semantic Image Segmentation

Xia, Xiaoling; Lu, Qinyang; Gu, Xin

doi:10.1088/1742-6596/1213/2/022003

Cited by 9 publications

(3 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For semantic segmentation (i.e., pixel-wise classification) of the catheter balloon, the commonly used U-Net architecture ( 22 ) was employed for both its high-resolution localization of objects and global context understanding, owing to its encoder-decoder structure. Here a modified U-Net model using a ResNet-34 ( 23 ) encoder (i.e., a 34 layered Residual Neural Network) was employed as it was shown to provide higher recall when compared to the standard U-Net architecture with imbalanced datasets ( 24 ), as is the case in this study (i.e., the balloon represents <1% of the image). The U-Net consisted of pre-trained weights using the ImageNet dataset ( 25 ).…”

Section: Methodsmentioning

confidence: 99%

Automatic image-based tracking of gadolinium-filled balloon wedge catheters for MRI-guided cardiac catheterization using deep learning

Neofytou,

Kowalik,

Vidya Shankar

et al. 2023

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

IntroductionMagnetic Resonance Imaging (MRI) is a promising alternative to standard x-ray fluoroscopy for the guidance of cardiac catheterization procedures as it enables soft tissue visualization, avoids ionizing radiation and provides improved hemodynamic data. MRI-guided cardiac catheterization procedures currently require frequent manual tracking of the imaging plane during navigation to follow the tip of a gadolinium-filled balloon wedge catheter, which unnecessarily prolongs and complicates the procedures. Therefore, real-time automatic image-based detection of the catheter balloon has the potential to improve catheter visualization and navigation through automatic slice tracking.MethodsIn this study, an automatic, parameter-free, deep-learning-based post-processing pipeline was developed for real-time detection of the catheter balloon. A U-Net architecture with a ResNet-34 encoder was trained on semi-artificial images for the segmentation of the catheter balloon. Post-processing steps were implemented to guarantee a unique estimate of the catheter tip coordinates. This approach was evaluated retrospectively in 7 patients (6M and 1F, age = 7 ± 5 year) who underwent an MRI-guided right heart catheterization procedure with all images acquired in an orientation unseen during training.ResultsThe overall accuracy, specificity and sensitivity of the proposed catheter tracking strategy over all 7 patients were 98.4 ± 2.0%, 99.9 ± 0.2% and 95.4 ± 5.5%, respectively. The computation time of the deep-learning-based segmentation step was ∼10 ms/image, indicating its compatibility with real-time constraints.ConclusionDeep-learning-based catheter balloon tracking is feasible, accurate, parameter-free, and compatible with real-time conditions. Online integration of the technique and its evaluation in a larger patient cohort are now warranted to determine its benefit during MRI-guided cardiac catheterization.

show abstract

Section: Methodsmentioning

confidence: 99%

Automatic image-based tracking of gadolinium-filled balloon wedge catheters for MRI-guided cardiac catheterization using deep learning

Neofytou,

Kowalik,

Vidya Shankar

et al. 2023

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

show abstract

“…4, Agustus 2023, hlm. 899-908 kenyamanan bagi para peneliti di bidang terkait dengan metode yang diusulkan block konvolusi berdasarkan Fibonacci dan menghasilkan metode sementasi yang efektif dengan perbedaan yang signifikan antara gambar background dan foregroundnya (Xia et al, 2019).…”

Section: 2unclassified

Peningkatan Performa Ensemble Learning pada Segmentasi Semantik Gambar dengan Teknik Oversampling untuk Class Imbalance

Nugroho,

Soeleman,

Pramunendar

et al. 2023

JTIIK

View full text Add to dashboard Cite

Perkembangan teknologi dan gaya hidup manusia yang semakin tinggi menghasilkan data-data yang berlimpah. Data-data tersebut dapat berbentuk data yang terstruktur dan tidak terstruktur. Data gambar termasuk dalam data yang tidak terstruktur. Aktifitas dan objek yang terekam dalam suatu gambar beraneka ragam. Secara normal, mata manusia dapat dengan mudah membedakan antara foreground dan background dari suatu gambar, tetapi komputer membutuhkan pembelajaran dalam membedakan keduanya. Segmentasi gambar adalah salah satu bidang dalam computer vision yang membahas bagaimana cara komputer mempelajari dan mengenali segmen dari suatu gambar sesuai label yang ditentukan. Dalam kenyataannya banyak data yang mempunyai class atau label yang tidak seimbang, tentunya akan mempengaruhi tingkat akurasi dari suatu prediksi. Dalam riset ini membahas bagaimana meningkatkan akurasi segmentasi semantik gambar pada metode ensemble learning untuk menangani masalah data yang tidak seimbang dalam segmentasi gambar. Teknik yang digunakan adalah sintetis oversampling sehingga menghasilkan data yang seimbang dan akurasi yang tinggi. Metode ensemble learning yang digunakan adalah Random Forest dan Light Gradien Boosting Machine (LGBM). Dengan menggunakan dataset Penn-Fudan Database for Pedestrian yang mengandung imbalanced class. Penggunaan teknik sintetis oversampling dapat memperbaikki tingkat akurasi pada class minoritas. Pada algoritma random forest mengalami peningkatan akurasi sebesar 37 % sedangkan pada algoritma LGBM meningkat sebesar 41 %. AbstractThe development of technology and the increasingly high lifestyle of humans produce abundant data. These data can be in the form of structured and unstructured data. Image data is included in unstructured data. The activities and objects recorded in a picture are varied. Normally, the human eye can easily distinguish between the foreground and background of an image, but computers need learning to distinguish between the two. Image segmentation is one of the fields in computer vision that discusses how computers learn and recognize segments of an image according to specified labels. In reality, a lot of data has unbalanced classes or labels, of course, it will affect the accuracy of a prediction. This research discusses how to improve the accuracy of image semantic segmentation in the ensemble learning method to deal with the problem of unbalanced data in image segmentation. The technique used is synthetic oversampling so as to produce balanced data and high accuracy. The ensemble learning methods used are Random Forest and Light Gradient Boosting Machine (LGBM). By using the Penn-Fudan Database for Pedestrian dataset which contains a imbalanced class. The use of synthetic oversampling techniques can improve the level of accuracy in minority classes. The random forest algorithm experienced an increase in accuracy by 37% while the LGBM algorithm increased by 41%.

show abstract

“…The first method implies using a weighted loss function during training (e.g., Weighted Binary Cross Entropy) that assigns a larger weight to samples containing buildings and, therefore, induces stronger changes in the net parameters when a building is being processed. The second method [33] suggests training the model only on positive examples, i.e., patches containing more than a pre-set number or percentage of building pixels in our case. This second approach was selected because it is expected not to affect the generalization capabilities of the network.…”

Section: Training Procedures 431 Data Processingmentioning

confidence: 99%

Integrating EfficientNet into an HAFNet Structure for Building Mapping in High-Resolution Optical Earth Observation Data

Ferrari

Dell’Acqua

Zhang³

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Automated extraction of buildings from Earth observation (EO) data is important for various applications, including updating of maps, risk assessment, urban planning, and policy-making. Combining data from different sensors, such as high-resolution multispectral images (HRI) and light detection and ranging (LiDAR) data, has shown great potential in building extraction. Deep learning (DL) is increasingly used in multi-modal data fusion and urban object extraction. However, DL-based multi-modal fusion networks may under-perform due to insufficient learning of “joint features” from multiple sources and oversimplified approaches to fusing multi-modal features. Recently, a hybrid attention-aware fusion network (HAFNet) has been proposed for building extraction from a dataset, including co-located Very-High-Resolution (VHR) optical images and light detection and ranging (LiDAR) joint data. The system reported good performances thanks to the adaptivity of the attention mechanism to the features of the information content of the three streams but suffered from model over-parametrization, which inevitably leads to long training times and heavy computational load. In this paper, the authors propose a restructuring of the scheme, which involved replacing VGG-16-like encoders with the recently proposed EfficientNet, whose advantages counteract exactly the issues found with the HAFNet scheme. The novel configuration was tested on multiple benchmark datasets, reporting great improvements in terms of processing times, and also in terms of accuracy. The new scheme, called HAFNetE (HAFNet with EfficientNet integration), appears indeed capable of achieving good results with less parameters, translating into better computational efficiency. Based on these findings, we can conclude that, given the current advancements in single-thread schemes, the classical multi-thread HAFNet scheme could be effectively transformed by the HAFNetE scheme by replacing VGG-16 with EfficientNet blocks on each single thread. The remarkable reduction achieved in computational requirements moves the system one step closer to on-board implementation in a possible, future “urban mapping” satellite constellation.

show abstract

Exploring An Easy Way for Imbalanced Data Sets in Semantic Image Segmentation

Cited by 9 publications

References 5 publications

Automatic image-based tracking of gadolinium-filled balloon wedge catheters for MRI-guided cardiac catheterization using deep learning

Automatic image-based tracking of gadolinium-filled balloon wedge catheters for MRI-guided cardiac catheterization using deep learning

Peningkatan Performa Ensemble Learning pada Segmentasi Semantik Gambar dengan Teknik Oversampling untuk Class Imbalance

Integrating EfficientNet into an HAFNet Structure for Building Mapping in High-Resolution Optical Earth Observation Data

Contact Info

Product

Resources

About