Unbiased Risk Estimate Research Articles

Risk of Syndrome-Associated Cancers Among First-Degree Relatives of Patients With Pancreatic Ductal Adenocarcinoma With Pathogenic or Likely Pathogenic Germline Variants

Increased cancer risk in first-degree relatives of probands with pancreatic ductal adenocarcinoma (PDAC probands) who carry pathogenic or likely pathogenic germline variants (PGVs) in cancer syndrome-associated genes encourages cascade genetic testing. To date, unbiased risk estimates for the development of cancers on a gene-specific basis have not been assessed. To quantify the risk of development of PDAC and extra-PDAC among first-degree relatives of PDAC probands who carry a PGV in 1 of 9 cancer syndrome-associated genes-ATM, BRCA1, BRCA2, PALB2, MLH1, MSH2, MSH6, PMS2, and CDKN2A. This case series focused on first-degree relatives of PDAC probands carrying PGVs in specific cancer syndrome-associated genes. The cohort comprised clinic-ascertained patients enrolled in the Mayo Clinic Biospecimen Resource for Pancreas Research registry with germline genetic testing. In total, 234 PDAC probands carrying PGVs were drawn from the prospective research registry of 4562 participants who had undergone genetic testing of cancer syndrome-associated genes. Demographic and cancer-related family histories were obtained by questionnaire. The data were collected from October 1, 2000, to December 31, 2021. For the PDAC probands, the genetic test results of the presence of PGVs in 9 cancer syndrome-associated genes were obtained by clinical testing. Cancers (ovary, breast, uterus or endometrial, colon, malignant melanoma, and pancreas) among first-degree relatives were reported by the probands. Standardized incidence ratios (SIRs) were used to estimate cancer risks among first-degree relatives of PDAC probands carrying a PGV. In total, 1670 first-degree relatives (mean [SD] age, 58.1 [17.8] years; 853 male [51.1%]) of 234 PDAC probands (mean [SD] age, 62.5 [10.1] years; 124 male [53.0%]; 219 [94.4%] White; 225 [98.7%] non-Hispanic or non-Latino]) were included in the study. There was a significantly increased risk of ovarian cancer in female first-degree relatives of probands who had variants in BRCA1 (SIR, 9.49; 95% CI, 3.06-22.14) and BRCA2 (SIR, 3.72; 95% CI, 1.36-8.11). Breast cancer risks were higher with BRCA2 variants (SIR, 2.62; 95% CI, 1.89-3.54). The risks of uterine or endometrial cancer (SIR, 6.53; 95% CI, 2.81-12.86) and colon cancer (SIR, 5.83; 95% CI, 3.70-8.75) were increased in first-degree relatives of probands who carried Lynch syndrome mismatch repair variants. Risk of PDAC was also increased for variants in ATM (SIR, 4.53; 95% CI, 2.69-7.16), BRCA2 (SIR, 3.45; 95% CI, 1.72-6.17), CDKN2A (SIR, 7.38; 95% CI, 3.18-14.54), and PALB2 (SIR, 5.39; 95% CI, 1.45-13.79). Melanoma risk was elevated for first-degree relatives of probands with CDKN2A variants (SIR, 7.47; 95% CI, 3.97-12.77). In this case series, the presence of PGVs in 9 cancer syndrome-associated genes in PDAC probands was found to be associated with increased risk of 6 types of cancers in first-degree relatives. These gene-specific PDAC and extra-PDAC cancer risks may provide justification for clinicians to counsel first-degree relatives about the relevance and importance of genetic cascade testing, with the goal of higher uptake of testing.

ENSURE: A General Approach for Unsupervised Training of Deep Image Reconstruction Algorithms.

Image reconstruction using deep learning algorithms offers improved reconstruction quality and lower reconstruction time than classical compressed sensing and model-based algorithms. Unfortunately, clean and fully sampled ground-truth data to train the deep networks is often unavailable in several applications, restricting the applicability of the above methods. We introduce a novel metric termed the ENsemble Stein's Unbiased Risk Estimate (ENSURE) framework, which can be used to train deep image reconstruction algorithms without fully sampled and noise-free images. The proposed framework is the generalization of the classical SURE and GSURE formulation to the setting where the images are sampled by different measurement operators, chosen randomly from a set. We evaluate the expectation of the GSURE loss functions over the sampling patterns to obtain the ENSURE loss function. We show that this loss is an unbiased estimate for the true mean-square error, which offers a better alternative to GSURE, which only offers an unbiased estimate for the projected error. Our experiments show that the networks trained with this loss function can offer reconstructions comparable to the supervised setting. While we demonstrate this framework in the context of MR image recovery, the ENSURE framework is generally applicable to arbitrary inverse problems.

Robust compressed sensing MRI based on combined nonconvex regularization

Compressed sensing (CS) is well known method for solving fast reconstruction of magnetic resonance (MR) images. However, the quality of the reconstructed images is sensitive to the noise. The reconstructed results often suffer from blurry or jaggy artifacts, which seriously affects the analysis of the medical image. To achieve robustness reconstruction result, in this paper, the combined nonconvex regularization model is proposed to enhance the details recovery and reduce jaggy artifacts. In the proposed framework, the bias-based sharpness enhancement prior is integrated into smoothed L0 (SL0) model, which can make SL0 reconstruction method more robust to noise. To solve the nonconvex optimization problem, a two-cycle iterative algorithm is developed to solve it. In addition, the improved Stein Unbiased Risk Estimator (SURE) is developed to automatically tune the shape parameter that to be effective for a variety of noise levels. The experimental results on MR images consistently demonstrate the excellent performance of the proposed method in both the objective quality metrics and visual inspections.

Beamspace Channel Estimation for Wideband Millimeter-Wave MIMO: A Model-Driven Unsupervised Learning Approach

Millimeter-wave (mmWave) communications have been one of the promising technologies for future wireless networks that integrate a wide range of data-demanding applications. To compensate for the large channel attenuation in mmWave band and avoid high hardware cost, a lens-based beamspace massive multiple-input multiple-output (MIMO) system is considered. However, the spatial-wideband effect in wideband mmWave systems makes channel estimation very challenging, especially when the receiver is equipped with a limited number of radio-frequency (RF) chains. Furthermore, the real channel data cannot be obtained before the mmWave system is used in a new environment, which makes it impossible to train a deep learning (DL)-based channel estimator using real data set beforehand. To solve the problem, we propose a model-driven unsupervised learning network, named learned denoising-based generalized expectation consistent (LDGEC) signal recovery network. By utilizing the Stein’s unbiased risk estimator loss, the LDGEC network can be trained only with limited measurements corresponding to the pilot symbols, instead of the real channel data. Even if designed for unsupervised learning, the LDGEC network can be supervisingly trained with the real channel via the denoiser-by-denoiser way. The numerical results demonstrate that the LDGEC-based channel estimator significantly outperforms state-of-the-art compressive sensing-based algorithms when the receiver is equipped with a small number of RF chains and low-resolution ADCs.

Deep external and internal learning for noisy compressive sensing

Reconstructing natural image from its corresponding compressive sensing (CS) measurements is an ill-posed problem. Learning accurate prior of desirable image is essential to solve this inverse problem in high quality, especially for CS reconstruction from noisy measurements. The existing learning-based methods cannot effectively simulate whole potential noise in the testing data during external learning, and cannot effectively exploit the information from internal testing data. In this paper, we present an effective convolutional neural network (CNN) based method for CS reconstruction from noisy measurements, which learns the deep prior from an external dataset and internal noisy testing data with Stein’s unbiased risk estimator (SURE). Specifically, we first pre-train an arbitrary CNN for CS reconstruction with an external dataset to learn a common prior. Then, we utilize meta learning to find a generic initial parameter that is suitable for fast internal learning and adaption to various noisy measurements. Finally, we customize the learned network to learn a specific prior for each internal testing data under Gaussian noise or more general mixed Poisson-Gaussian noise. Experimental results show that the proposed method outperforms the state-of-the-art methods under both comprehensively quantitative metrics and perceptive quality.

Automatic parameter tuning for plug-and-play algorithms using generalized cross validation and Stein's unbiased risk estimation for linear inverse problems in computational imaging

We propose two automatic parameter tuning methods for Plug-and-Play (PnP) algorithms that use CNN denoisers. We focus on linear inverse problems and propose an iterative algorithm to calculate generalized cross-validation (GCV) and Stein’s unbiased risk estimator (SURE) functions for a half-quadratic splitting-based PnP (PnP-HQS) algorithm that uses a state-of- the-art CNN denoiser. The proposed methods leverage forward mode automatic differentiation to calculate the GCV and SURE functions and tune the parameters of a PnP-HQS algorithm automatically by minimizing the GCV and SURE functions using grid search. Because linear inverse problems appear frequently in computational imaging, the proposed methods can be applied in various domains. Furthermore, because the proposed methods rely on GCV and SURE functions, they do not require access to the ground truth image and do not require collecting an additional training dataset, which is highly desirable for imaging applications for which acquiring data is costly and time-consuming. We evaluate the performance of the proposed methods on deblurring and MRI experiments and show that the GCV-based proposed method achieves comparable performance to that of the oracle tuning method that adjusts the parameters by maximizing the structural similarity index between the ground truth image and the output of the PnP algorithm. We also show that the SURE-based proposed method often leads to worse performance compared to the GCV-based proposed method.

Combining observational and experimental datasets using shrinkage estimators.

We consider the problem of combining data from observational and experimental sources to draw causal conclusions. To derive combined estimators with desirable properties, we extend results from the Stein shrinkage literature. Our contributions are threefold. First, we propose a generic procedure for deriving shrinkage estimators in this setting, making use of a generalized unbiased risk estimate. Second, we develop two new estimators, prove finite sample conditions under which they have lower risk than an estimator using only experimental data, and show that each achieves a notion of asymptotic optimality. Third, we draw connections between our approach and results in sensitivity analysis, including proposing a method for evaluating the feasibility of ourestimators.

Performance Enhancement of INS and UWB Fusion Positioning Method Based on Two-Level Error Model

In GNSS-denied environments, especially when losing measurement sensor data, inertial navigation system (INS) accuracy is critical to the precise positioning of vehicles, and an accurate INS error compensation model is the most effective way to improve INS accuracy. To this end, a two-level error model is proposed, which comprehensively utilizes the mechanism error model and propagation error model. Based on this model, the INS and ultra-wideband (UWB) fusion positioning method is derived relying on the extended Kalman filter (EKF) method. To further improve accuracy, the data prefiltering algorithm of the wavelet shrinkage method based on Stein's unbiased risk estimate-Shrink (SURE-Shrink) threshold is summarized for raw inertial measurement unit (IMU) data. The experimental results show that by employing the SURE-Shrink wavelet denoising method, positioning accuracy is improved by 76.6%; by applying the two-level error model, the accuracy is further improved by 84.3%. More importantly, at the point when the vehicle motion state changes, adopting the two-level error model can provide higher computational stability and less fluctuation in trajectory curves.

A Self-Supervised Denoising Method Based on Deep Noise Estimation

Seismic random noise attenuation is an essential procedure when processing seismic data. Due to various acquisition environments and complex geological conditions, random noise in seismic field data exhibits spatiotemporal levels, significantly increasing the difficulty of extracting seismic signals. The deep learning (DL) methods have shown excellent performances for seismic denoising. However, most existing discriminative DL methods cannot fit field data with noise levels and signal structures that differ from the training set. To tackle the unmatching challenge, we propose a self-supervised deep denoising model named noise estimation-based convolution neural network (NE-CNN), which contains a multiscale denoising module as a pretrained model and a dual-path noise estimation module that estimates the noise level of each data patch with a generalized Gaussian distribution and gray-level co-occurrence matrix (GLCM). With Stein’s unbiased risk estimate (SURE), we can fine-tune the pretrained denoising module according to the estimated noise levels in a self-supervised style solely on data to be processed, leading to a boost of robustness to nonstationary seismic noise. In synthetic and field data tests, NE-CNN performed satisfactorily compared with discriminative DL methods in denoising effects on seismic data with nonstationary noise.

Learning With Proper Partial Labels.

Partial-label learning is a kind of weakly supervised learning with inexact labels, where for each training example, we are given a set of candidate labels instead of only one true label. Recently, various approaches on partial-label learning have been proposed under different generation models of candidate label sets. However, these methods require relatively strong distributional assumptions on the generation models. When the assumptions do not hold, the performance of the methods is not guaranteed theoretically. In this letter, we propose the notion of properness on partial labels. We show that this proper partial-label learning framework requires a weaker distributional assumption and includes many previous partial-label learning settings as special cases. We then derive a unified unbiased estimator of the classification risk. We prove that our estimator is risk consistent, and we also establish an estimation error bound. Finally, we validate the effectiveness of our algorithm through experiments.

Learning spectral windowing parameters for regularization using unbiased predictive risk and generalized cross validation techniques for multiple data sets

During the inversion of discrete linear systems noise in data can be amplified and result in meaningless solutions. To combat this effect, regularization is imposed during the inversion. The influence of provided prior information is controlled by non-negative regularization parameter(s). There are a number of methods used to select appropriate regularization parameters. New methods of unbiased risk estimation and generalized cross validation are derived for finding spectral windowing regularization parameters. These estimators are extended for finding the regularization parameters when multiple data sets with common system matrices are available. It is demonstrated that spectral windowing regularization parameters can be learned from these new estimators applied for multiple data and with multiple windows. The results demonstrate that these modified methods, which do not require the use of true data for learning regularization parameters, are effective and efficient, and perform comparably to a learning method based on estimating the parameters using true data. The theoretical developments are validated for the case of two dimensional image deblurring. The results verify that the obtained estimates of spectral windowing regularization parameters can be used effectively on validation data sets that are separate from the training data, and do not require known data.

UNBIASED THRESHOLDING RISK ESTIMATE WITH TWO THRESHOLD VALUES

UNBIASED THRESHOLDING RISK ESTIMATE WITH TWO THRESHOLD VALUES

VOLATILITY FORECASTING AFTER REMOVING THE NOISE: DOES IT REALLY PAY OFF?

The analysis in this paper represents the first attempt to examine whether the forecasting performance of one of the GARCH models can be improved after removing the noise from the stock market returns. Doing the analysis on eight developed stock markets helps in providing reliable results that can be generalized to more stock markets. This paper then examines the ability of GARCH models to forecast stock return volatility under a range of statistical forecasts. We specifically employ the multiscale-based approach, namely the maximum overlap discrete wavelet transformation, to remove the noisy data from the return series. Based on related theoretical literature, we assume that noise trading makes the forecasting performance more misleading, especially at the short-term horizon. Our particular interest then is whether wavelet de-noising of the data before estimation affects the ability of the models to provide accurate forecasts at a short investment horizon. To de-noise the data, we use soft thresholding and Stein’s Unbiased Risk Estimator to obtain the decomposition level-based threshold limit. Our key results demonstrate that de-noising returns improves the accuracy of volatility forecasts regardless of whether we used statistical metrics or tests of equal predictive accuracy. Moreover, in terms of a particular volatility model, the asymmetric GARCH approach tends to be preferred although this result is not universal. Indeed, the central result from our analysis is that the process of de-noising is more important than the specific model.

Adaptive and efficient estimation in the Gaussian sequence model

This paper deals with the problem of estimating the mean of an infinite-dimensional Gaussian vector by the principle of minimizing unbiased risk estimation (URE). The aim is to obtain an adaptive estimator to mimic the oracle with the smallest squared L2 loss/risk in a given class of linear estimators Λ. It is noticed that there is an essential balance between enlarging Λ for more efficient oracle risk and contracting Λ for the adaptivity to the oracle. This paper proves that this balance can be well achieved by minimizing URE in a list of intermediate classes between the projection and monotone classes. The resulting estimators are adaptive on the nonparametric space and have more efficient risks than the projection estimators under certain conditions.

Hyperparameter selection for Discrete Mumford–Shah

This work focuses on a parameter-free joint piecewise smooth image denoising and contour detection. Formulated as the minimization of a discrete Mumford-Shah functional and estimated via a theoretically grounded alternating minimization scheme, the bottleneck of such a variational approach lies in the need to fine-tune their hyperparameters, while not having access to ground truth data. To that aim, a Stein-like strategy providing optimal hyperparameters is designed, based on the minimization of an unbiased estimate of the quadratic risk. Efficient and automated minimization of the estimate of the risk crucially relies on an unbiased estimate of the gradient of the risk with respect to hyperparameters. Its practical implementation is performed using a forward differentiation of the alternating scheme minimizing the Mumford-Shah functional, requiring exact differentiation of the proximity operators involved. Intensive numerical experiments are performed on synthetic images with different geometry and noise levels, assessing the accuracy and the robustness of the proposed procedure. The resulting parameter-free piecewise-smooth estimation and contour detection procedure, not requiring prior image processing expertise nor annotated data, can then be applied to real-world images.

Abstract A033: Gene-specific risk of syndrome-associated cancers in first-degree relatives of pancreatic cancer patients with pathogenic/likely pathogenic variants

Abstract Increased pancreatic cancer (PC) risk in first-degree relatives (FDRs) of PC patients with germline pathogenic/likely pathogenic variants (PLPV) is a motivator for cascade genetic testing. To date, unbiased FDR risk estimates for gene-specific risks of genetic cancer syndrome-associated cancers are unknown. These estimates would enhance genetic counseling of at-risk FDRs in families where the PC proband has been found to carry a PLPV. Study Aim: To objectively quantify gene-specific risks of six relevant cancers among FDRs of PC patients with germline PLPV in ATM, BRCA1, BRCA2, CDKN2A, MLH1, MSH2, MSH6, PALB2, and PMS2. Methods: In the prospective, clinic-based Mayo Clinic Biospecimen Resource for Pancreas Research registry, 4,562 PC patients had previously undergone germline genetic testing for pancreatic cancer-associated genes through either research studies or clinical testing. Of these, 234 PC probands were found to carry PLPV among 9 genes of interest and had provided detailed demographic and cancer data on their FDRs by questionnaire. We focused on six cancer types (ovary, breast, uterus, pancreas, colon, and malignant melanoma) in FDRs as reported by the probands. Standardized incidence ratios estimated risk by gene of six cancers among FDRs of PC patients carrying PLPV. Results: 1,670 FDRs (mean age 58.1+17.8SD; 48.9% female) were included in the study. We found significantly increased risk of ovarian cancer in female FDRs of PC probands who carry PLPV in BRCA1 (SIR 9.49, 95%CI:3.06-22.14) or BRCA2 (3.72, 95%CI:1.36-8.11), and breast cancer risks were higher with BRCA2 (2.62, 95%CI:1.89-3.54). Uterine cancer risk was increased in FDRs of PC probands who carry PLPV for Lynch Syndrome mismatch repair (MMR) (6.53, 95%CI:2.81-12.86). PC risk was also increased (ATM 4.53, 95% CI:2.69-7.16; BRCA2 3.45, 95%CI:1.72-6.17; CDKN2A 7.38, 95%CI:3.18-14.54; PALB2 5.39, 95%CI:1.45-13.79). Increased colon cancer risk was observed in FDRs of probands who carried MMR PLPV (5.83, 95%CI:3.70-8.75), while melanoma risk was elevated for FDRs of probands with CDKN2A PLPV (7.47, 95%CI:3.97-12.77). Conclusion: PLPV in nine syndrome-associated genes in PC probands are associated with increased risk of six cancers in FDRs in PC families. The findings underscore the importance of risk estimation of various other cancers in PC families for screening, early detection, and intervention. The gene-specific quantified cancer risks for FDRs of PC patients with syndrome-associated germline PLPV can inform counseling regarding extra-PC risks when considering cascade genetic testing. Citation Format: Xuan Chen, Kari Rabe, Margaret Meyer, Jennnifer Kemppainen, Masayasu Horibe, Shruti Chandra, Shounak Majumder, Gloria Petersen. Gene-specific risk of syndrome-associated cancers in first-degree relatives of pancreatic cancer patients with pathogenic/likely pathogenic variants [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer; 2022 Sep 13-16; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2022;82(22 Suppl):Abstract nr A033.

Relaxing the Gaussian assumption in shrinkage and SURE in high dimension

Shrinkage estimation is a fundamental tool of modern statistics, pioneered by Charles Stein upon his discovery of the famous paradox involving the multivariate Gaussian. A large portion of the subsequent literature only considers the efficiency of shrinkage, and that of an associated procedure known as Stein’s Unbiased Risk Estimate, or SURE, in the Gaussian setting of that original work. We investigate what extensions to the domain of validity of shrinkage and SURE can be made away from the Gaussian through the use of tools developed in the probabilistic area now known as Stein’s method. We show that shrinkage is efficient away from the Gaussian under very mild conditions on the distribution of the noise. SURE is also proved to be adaptive under similar assumptions, and in particular in a way that retains the classical asymptotics of Pinsker’s theorem. Notably, shrinkage and SURE are shown to be efficient under mild distributional assumptions, and particularly for general isotropic log-concave measures.

Bridging between Soft and Hard Thresholding by Scaling

In this article, we developed and analyzed a thresholding method in which soft thresholding estimators are independently expanded by empirical scaling values. The scaling values have a common hyper-parameter that is an order of expansion of an ideal scaling value that achieves hard thresholding. We simply call this estimator a scaled soft thresholding estimator. The scaled soft thresholding is a general method that includes the soft thresholding and non-negative garrote as special cases and gives an another derivation of adaptive LASSO. We then derived the degree of freedom of the scaled soft thresholding by means of the Stein's unbiased risk estimate and found that it is decomposed into the degree of freedom of soft thresholding and the reminder connecting to hard thresholding. In this meaning, the scaled soft thresholding gives a natural bridge between soft and hard thresholding methods. Since the degree of freedom represents the degree of over-fitting, this result implies that there are two sources of over-fitting in the scaled soft thresholding. The first source originated from soft thresholding is determined by the number of un-removed coefficients and is a natural measure of the degree of over-fitting. We analyzed the second source in a particular case of the scaled soft thresholding by referring a known result for hard thresholding. We then found that, in a sparse, large sample and non-parametric setting, the second source is largely determined by coefficient estimates whose true values are zeros and has an influence on over-fitting when threshold levels are around noise levels in those coefficient estimates. In a simple numerical example, these theoretical implications has well explained the behavior of the degree of freedom. Moreover, based on the results here and some known facts, we explained the behaviors of risks of soft, hard and scaled soft thresholding methods.

Degrees of freedom for off-the-grid sparse estimation

The degrees of freedom quantify the number of parameters of an estimator and is central to various risk minimization procedures. Its computation and properties are well understood when dealing with finite dimensional linear models, possibly regularized using sparsity through the celebrated Lasso estimator. However, for some applications (such as super-resolution in imaging or training multi-layer perceptrons with a single hidden layer) it makes sense to rather consider “continuous” methods. For these applications, these methods avoid the discretization of the parameter space and lead to more efficient numerical solvers. Training these continuous models with a sparsity inducing prior can be achieved by solving a convex optimization problem over the infinite dimensional space of measures, which is often called the Beurling Lasso (Blasso), and is the continuous counterpart of the Lasso. Previous works (Ann. Statist. 35 (2007) 2173–2192; Statist. Sinica 23 (2013) 809–828) show that the size of the smallest solution support is an unbiased estimator of the degrees of freedom of the Lasso. Our main contribution is a proof of a continuous counterpart to this result for the Blasso. In contrast to the Lasso, our new formula shows that the empirical degrees of freedom of the Blasso is not proportional to the size of the support. Our second contribution is a detailed study of the case of sampling Fourier coefficients in 1D, which corresponds to a super-resolution problem. We show that our formula for the unbiased estimation of the degrees of freedom is valid outside of a set of measure zero of observations, which in turn justifies its use to compute an unbiased estimator of the prediction risk using the Stein Unbiased Risk Estimator (SURE). We also report numerical results for both the case of Fourier sampling and the learning of a multilayers perceptron with a single hidden layer.

Detecting pine wilt disease at the pixel level from high spatial and spectral resolution UAV-borne imagery in complex forest landscapes using deep one-class classification

Pine wilt disease (PWD) poses a serious threat to the worldwide pine forest resources. Unmanned aerial vehicle (UAV) remote sensing has been widely used for PWD control, due to its flexibility and efficiency. Although pixel-level detection can obtain fine detection boundaries, there have been few related works in complex scenes because of the difficulty of setting a preset category system and the poor generalization. A preset category system establishes which categories are to be labeled, and is necessary for traditional pixel-level detection. However, the poor generalization leads to an obvious accuracy drop when detecting PWD in new scenes. In the proposed approach, to address the first issue, one-class classification (OCC) is introduced to detect diseased pixels, focusing only on the category of diseased pine trees. However, the numerous objects but low PWD pixel proportion makes the model optimization unbalanced, for which balanced unbiased detection risk estimation is proposed. To address the second issue, a novel model consisting of three-dimensional (3D) convolutional layers and transformer blocks is proposed to extract more robust features. A novel PWD detection framework based on deep OCC is finally proposed to achieve fine pixel-level PWD detection results. PWD detection experiments were conducted on eight UAV H2 (high spatial and spectral resolution) image strips. In total, 300 PWD samples from strip 1 (accounting for roughly 0.009 % of the total pixels) and 400 unlabeled pixels formed the training set. The test experiments were conducted in the remaining seven strips to validate the model generalization. Satisfactory quantitative results (F1-score greater than 0.9) were obtained for all the test strips. The results indicate that the proposed method has a powerful ability to detect PWD in pine trees, even when the PWD proportion is low, and shows better model generalization than the traditional pixel-level detection methods.