Lung nodule detection in chest X-rays using synthetic ground-truth data comparing CNN-based diagnosis to human performance

Schultheiß, Manuel; Schmette, Philipp; Bodden, Jannis; Aichele, Juliane; Müller-Leisse, Christina; Gassert, Felix G.; Gassert, Florian T.; Gawlitza, Joshua; Hofmann, Felix; Sasse, Daniel; Schacky, Claudio E. von; Ziegelmayer, Sebastian; Marco, F. De; Renger, Bernhard; Makowski, Marcus R.; Pfeiffer, Franz; Pfeiffer, Daniela

doi:10.1038/s41598-021-94750-z

Cited by 21 publications

(11 citation statements)

References 49 publications

(26 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As the 3D CT scans for training X-ray image generation were from different patients compared with the real X-rays and the lesion annotations were performed by different expert clinicians, there is an inconsistency in the lesion appearance between training data and real X-ray data, which potentially causes the performance deterioration. Similar effects have previously been reported for related tasks, such as lung nodule detection 29 and thoracic disease classification 30 . The results suggest that SyntheX is capable of handling soft tissue-based tasks, such as COVID-19 lesion segmentation.…”

Section: Covid-19 Lesion Segmentationsupporting

confidence: 85%

Synthetic data accelerates the development of generalizable learning-based algorithms for X-ray image analysis

Gao

Killeen

et al. 2023

Nat Mach Intell

View full text Add to dashboard Cite

Artificial intelligence (AI) now enables automated interpretation of medical images. However, AI's potential use for interventional image analysis remains largely untapped. This is because the post hoc analysis of data collected during live procedures has fundamental and practical limitations, including ethical considerations, expense, scalability, data integrity and a lack of ground truth. Here we demonstrate that creating realistic simulated images from human models is a viable alternative and complement to large-scale in situ data collection. We show that training AI image analysis models on realistically synthesized data, combined with contemporary domain generalization techniques, results in machine learning models that on real data perform comparably to models trained on a precisely matched real data training set. We find that our model transfer paradigm for X-ray image analysis, which we refer to as SyntheX, can even outperform real-data-trained models due to the effectiveness of training on a larger dataset. SyntheX provides an opportunity to markedly accelerate the conception, design and evaluation of X-ray-based intelligent systems. In addition, SyntheX provides the opportunity to test novel instrumentation, design complementary surgical approaches, and envision novel techniques that improve outcomes, save time or mitigate human error, free from the ethical and practical considerations of live human data collection.Advances in robotics and artificial intelligence (AI) are bringing autonomous surgical systems closer to reality. However, developing the AI backbones of such systems currently depends on collecting training data during routine surgeries. This remains one of the largest barriers to widespread use of AI systems in interventional clinical settings, versus triage or diagnostic settings, as the acquisition and annotation of interventional data is time intensive and costly. Furthermore, while this approach can contribute to the automation or streamlining of existing surgical workflows, robotic and autonomous systems promise even more substantial advances: novel and super-human techniques that improve outcomes, save time or mitigate human error. This is perhaps the most exciting frontier of computer-assisted intervention research.Conventional approaches for curating data for AI development (that is, sourcing it retroactively from clinical practice) are insufficient for training AI models that benefit interventions that use novel instrumentation, different access points or more flexible imaging. This is because they are, by definition, incompatible with contemporary clinical practices and such data do not emerge from routine care. Furthermore, these novel systems are not readily approved, and thus not easily or quickly introduced into clinical practice. Ex vivo experimentation

show abstract

Section: Covid-19 Lesion Segmentationsupporting

confidence: 85%

Synthetic data accelerates the development of generalizable learning-based algorithms for X-ray image analysis

Gao

Killeen

et al. 2023

Nat Mach Intell

View full text Add to dashboard Cite

show abstract

“…Given the small remaining dataset size and inspired by Schultheiss et al., 32 we synthesized a larger amount of CT volumes with nodules that matched both the volume and contrast standards by inserting nodules of those high‐visibility patients into CT volumes of other low‐visibility patients. The procedure is illustrated in Figure 2.…”

Section: Methodsmentioning

confidence: 99%

“…Given the small remaining dataset size and inspired by Schultheiss et al, 32 we synthesized a larger amount of CT volumes with nodules that matched both the volume and contrast standards by inserting nodules of those high-visibility patients into CT volumes of other low-visibility patients. The procedure is illustrated in Upon insertion, nodules were randomly scaled by a ratio within 0.9 and 1.2, randomly rotated within 0 • to 360 • in all three axes, and inserted at a randomly sampled location within the lung mask of the selected CT volume.…”

Section: Data Augmentationmentioning

confidence: 99%

Simultaneous object detection and segmentation for patient‐specific markerless lung tumor tracking in simulated radiographs with deep learning

Huang,

Kurz,

Freislederer

et al. 2023

Medical Physics

View full text Add to dashboard Cite

BackgroundReal‐time tumor tracking is one motion management method to address motion‐induced uncertainty. To date, fiducial markers are often required to reliably track lung tumors with X‐ray imaging, which carries risks of complications and leads to prolonged treatment time. A markerless tracking approach is thus desirable. Deep learning‐based approaches have shown promise for markerless tracking, but systematic evaluation and procedures to investigate applicability in individual cases are missing. Moreover, few efforts have been made to provide bounding box prediction and mask segmentation simultaneously, which could allow either rigid or deformable multi‐leaf collimator tracking.PurposeThe purpose of this study was to implement a deep learning‐based markerless lung tumor tracking model exploiting patient‐specific training which outputs both a bounding box and a mask segmentation simultaneously. We also aimed to compare the two kinds of predictions and to implement a specific procedure to understand the feasibility of markerless tracking on individual cases.MethodsWe first trained a Retina U‐Net baseline model on digitally reconstructed radiographs (DRRs) generated from a public dataset containing 875 CT scans and corresponding lung nodule annotations. Afterwards, we used an independent cohort of 97 lung patients to develop a patient‐specific refinement procedure. In order to determine the optimal hyperparameters for automatic patient‐specific training, we selected 13 patients for validation where the baseline model predicted a bounding box on planning CT (PCT)‐DRR with intersection over union (IoU) with the ground‐truth higher than 0.7. The final test set contained the remaining 84 patients with varying PCT‐DRR IoU. For each testing patient, the baseline model was refined on the PCT‐DRR to generate a patient‐specific model, which was then tested on a separate 10‐phase 4DCT‐DRR to mimic the intrafraction motion during treatment. A template matching algorithm served as benchmark model. The testing results were evaluated by four metrics: the center of mass (COM) error and the Dice similarity coefficient (DSC) for segmentation masks, and the center of box (COB) error and the DSC for bounding box detections. Performance was compared to the benchmark model including statistical testing for significance.ResultsA PCT‐DRR IoU value of 0.2 was shown to be the threshold dividing inconsistent (68%) and consistent (100%) success (defined as mean bounding box DSC > 0.6) of PS models on 4DCT‐DRRs. Thirty‐seven out of the eighty‐four testing cases had a PCT‐DRR IoU above 0.2. For these 37 cases, the mean COM error was 2.6 mm, the mean segmentation DSC was 0.78, the mean COB error was 2.7 mm, and the mean box DSC was 0.83. Including the validation cases, the model was applicable to 50 out of 97 patients when using the PCT‐DRR IoU threshold of 0.2. The inference time per frame was 170 ms. The model outperformed the benchmark model on all metrics, and the comparison was significant (p < 0.001) over the 37 PCT‐DRR IoU > 0.2 cases, but not over the undifferentiated 84 testing cases.ConclusionsThe implemented patient‐specific refinement approach based on a pre‐trained baseline model was shown to be applicable to markerless tumor tracking in simulated radiographs for lung cases.

show abstract

“…Computed Tomography (CT) and Positron Emission Tomography (PET) are two of the most popular imaging techniques that may employ advanced X-ray equipment to scan the body and deliver detailed information regarding cancer-related activities. The inspection and detection of lung cancer depend greatly on the CT scan [10].…”

Section: Introductionmentioning

confidence: 99%

Detection & Quantification of Lung Nodules Using 3D CT images

Memon

Jawaid

Talpur

2023

IJIST

View full text Add to dashboard Cite

In computer vision image detection and quantification play an important role. Image Detection and quantification is the process of identifying nodule position and the amount of covered area. The dataset which we have used for this research contains 3D CT lung images. In our proposed work we have taken 3D images and those are high-resolution images. We have compared the accuracy of the existing mask and our segmented images. The segmentation method that we have applied to these images is Sparse Field Method localized region-based segmentation and for Nodule detection, I have used ray projection. The ray projection method is efficient for making the point more visible by its x, y, and z components. like a parametric equation where the line crossing through a targeted point by that nodule is more dominated. The Frangi filter was to give a geometric shape to the nodule and we got 90% accurate detection. The high mortality rate associated with lung cancer makes it imperative that it be detected at an early stage. The application of computerized image processing methods has the potential to improve both the efficiency and reliability of lung cancer screening. Computerized tomography (CT) pictures are frequently used in medical image processing because of their excellent resolution and low noise. Computer-aided detection systems, including preprocessing and segmentation methods, as well as data analysis approaches, have been investigated in this research for their potential use in the detection and diagnosis of lung cancer. The primary objective was to research cutting-edge methods for creating computational diagnostic tools to aid in the collection, processing, and interpretation of medical imaging data. Nonetheless, there are still areas that need more work, such as improving sensitivity, decreasing false positives, and optimizing the identification of each type of nodule, even those of varying size and form.

show abstract

Lung nodule detection in chest X-rays using synthetic ground-truth data comparing CNN-based diagnosis to human performance

Cited by 21 publications

References 49 publications

Synthetic data accelerates the development of generalizable learning-based algorithms for X-ray image analysis

Synthetic data accelerates the development of generalizable learning-based algorithms for X-ray image analysis

Simultaneous object detection and segmentation for patient‐specific markerless lung tumor tracking in simulated radiographs with deep learning

Detection & Quantification of Lung Nodules Using 3D CT images

Contact Info

Product

Resources

About