Interpretable active learning method cooperatively driven by cell image attributes

Abstract

The invention belongs to the technical field of interpretable active learning, and particularly relates to a cell image attribute collaborative driven interpretable active learning method, which comprises the following steps of: in an inspection process, acquiring an image from a blood cell smear by using a camera, constructing a newly acquired cell image data set D1, randomly extracting a small amount of samples from the data set D1 to construct an initial labeling set D2; performing category labeling on the D2 and completing model pre-training based on the D2, performing standardization and preprocessing on the remaining unlabeled data in the D1, dividing the remaining unlabeled data into a plurality of batches according to the number to form a candidate unlabeled pool D3, sequentially inputting the data in the D3 into a model according to a preset batch, and for each batch of input data, adopting a class imbalance adaptive sample screening mechanism CASS to obtain a candidate unlabeled pool D3; the method comprises the following steps: comprehensively utilizing three types of indexes of uncertainty, representativeness and diversity under current weight configuration, introducing attribute information, carrying out weighted calculation in combination with a type frequency and an effective sample number, and screening to obtain candidate high-value samples of the batch;

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of explainable active learning, and particularly relates to an explainable active learning method driven by cell image class attribute cooperation. BACKGROUND

[0002] In the intelligent application of cell morphology examination, although active learning technology has been used to reduce labeling costs, it still faces core technical bottlenecks. Traditional methods mainly focus on a single screening indicator and fail to fully combine cell morphology attributes, and are not good at dealing with the class imbalance problem commonly seen in clinical practice, which leads to rare pathological cell samples being easily ignored and weakens the model's ability to recognize minority classes, making it difficult to meet the accuracy requirements of clinical examination. At the same time, existing sample screening strategies lack dynamic adaptability and fail to adjust and optimize the direction at different stages of model training, further limiting the learning efficiency and model generalization performance.

[0003] In addition, mainstream active learning methods are mostly black box models, lacking explainability support for sample screening logic. Clinical examination experts cannot know the specific basis for the model's determination of high-value samples, making it difficult to effectively verify and trust the results, which seriously hinders the application of the technology in medical scenarios. At the same time, there is a lack of attribute semantic stability maintenance mechanism in the model training process, which easily leads to attribute drift, resulting in insufficient coherence in sample screening and model updating in the subsequent learning process, affecting the long-term optimization effect of the system. SUMMARY

[0004] The purpose of the present application is to provide a cell image class attribute cooperation driven explainable active learning method to solve the above technical problems.

[0005] Therefore, the present application provides a cell image class attribute cooperation driven explainable active learning method, comprising the following steps: Step 1: During the examination process, use a camera to obtain images from a blood smear, and construct a newly collected cell image dataset D1; Step 2: Randomly extract a small number of samples from the dataset D1 to construct an initial labeled set D2, label the classes of D2 and complete the pre-training of the model based on it; after standardizing and preprocessing the remaining unlabeled data in D1, divide them into several batches according to the number to form a candidate unlabeled pool D3; Step 3: Input the data in D3 into the model in batches according to the preset batch size, for each batch of input data, use the class imbalance adaptive sample screening mechanism CASS to comprehensively utilize uncertainty, representativeness and diversity under the current weight configuration, introduce attribute information, combine class frequency and effective sample number for weighted calculation, and screen out the candidate high-value samples of this batch; Step four: The weights of the four types of indicators, uncertainty, representativeness, diversity, and attribute information, are scheduled by the dynamic mixer AdaptiveMixer to realize the gradual switching of the Early, Middle, and Late stages; among them, the Early stage emphasizes representativeness and diversity to expand the sample coverage, the Middle stage keeps the weights of the four types of indicators relatively balanced, and the Late stage increases the weights of uncertainty and attributes; the weights output by the dynamic mixer are used to guide the sample selection strategy in the corresponding stage; Step five: After each batch of samples is selected, the selected new samples are used to fine-tune the model with a small number of steps, and the model is updated using the Exponential Moving Average (EMA) and light regularization; the classification accuracy acc, balanced accuracy bal_acc, macro-average F1 value macro-F1, and expected calibration error ECE indicators are monitored in the evaluation stage to ensure the stable performance of the model in terms of class balance and calibration; Step six: During active learning, the attribute information of each batch of selected sample sets is explained after the fact using the AIAL method, which explains the reasons why the samples are determined to be high-value based on cell morphology knowledge, and the explanation results are visualized and output; Step seven: The system optimized and explained by the above steps is used for a new round of blood cell morphology inspection, and steps one to six are repeated to realize the continuous evolution and dynamic optimization of the system.

[0006] Preferably, before model pre-training in step two, the backbone network and attribute network layer construction need to be completed, which specifically includes: the backbone network includes a low-dimensional feature extractor L and a high-dimensional feature extractor H; the low-dimensional feature extractor L is composed of the first several residual blocks of the pre-trained model ResNeXt50, which is used to extract and output basic morphological features; The high-dimensional feature extractor H is composed of the last several residual blocks of the pre-trained model ResNeXt50, which is used to learn and output higher-level distinguishing features of cells; An attribute network layer ANL is introduced before the category classifier C to project the high-dimensional features output by the high-dimensional feature extractor H into the attribute space to obtain an attribute vector, and the cell morphology attribute dimensions corresponding to the attribute vector include cell size, nucleus-cytoplasm ratio, and particle density; The attribute network layer ANL is composed of two linear mappings, which is responsible for compressing and aligning the high-dimensional representation in the high-dimensional features to the attribute semantic space, so that the system obtains an interpretable attribute representation in addition to the classification output, forming a dual-view representation structure of classification prediction and attribute explanation.

[0007] Preferably, before model pre-training, the strategy and mixer initialization also need to be completed, which specifically includes: The modules required for the four indicators are instantiated, including the uncertainty strategy U, the representativeness strategy R, the diversity strategy D, and the attribute value channel A; The uncertainty policy U employs an entropy measure of how ambiguous the sample is located at the decision boundary; The representative policy R employs KNN-density, which measures the coverage of the sample to the overall distribution by calculating the Top-k cosine similarity mean and normalizing to [0, 1]; The diversity policy D employs K-Center strategy to encourage the scattered selection of feature space; The attribute value channel A is composed of attribute uncertainty, attribute novelty and attribute coverage gap; An adaptive mixer is constructed, and in the initialization stage, the mixer is provided with three-stage weights according to the preset definition. The adaptive switching mechanism automatically decides the weight configuration according to the batch performance consistency index ST. After active learning iteration sampling, the mixer loads the corresponding policy configuration according to the current stage it belongs to, and dynamically switches between Early-Middle-Late.

[0008] Preferably, the implementation mode of the adaptive switching mechanism in step four is as follows: the performance consistency index ST is used to comprehensively measure the performance improvement, consistency and calibration state of the current batch relative to the last batch. The calculation basis includes the overall prediction accuracy, class average recall rate, macro average F1 score and expected calibration error change on the validation set; The stage switching rule is: if the performance consistency index ST exceeds the switching threshold of the current stage for h consecutive times, then enter the next stage; If the performance consistency index ST significantly decreases below the rollback threshold, or a core performance index continuously degrades, then rollback from the high stage to the previous stage; The adaptive mixer outputs four-way indicators of the weight at each batch, which is directly used as the input of the class imbalance adaptive sample screening mechanism CASS of the corresponding stage, guiding the comprehensive evaluation, sorting and selection of candidate samples.

[0009] Preferably, the specific implementation process of the class imbalance adaptive sample screening mechanism CASS in step three includes: Class statistics and effective sample number weighting; The class frequency of the labeled sample is counted, the class sensitivity score is constructed based on the class frequency, the class sensitivity score is normalized and upper and lower limit truncated to obtain the final class weighting, which is used to improve the selection priority of the minority class sample; class sensitive re-labeling: For each candidate sample, the expected rarity is calculated by using the prediction distribution and the class weighting as a sample rarity adjustment factor to participate in the final sampling; class sensitive deduplication and class quota joint constraint: For the same category of candidate samples, the similarity threshold is set according to the category sensitive weighting to perform the deduplication strategy, and the deduplication strategy is not performed between different categories. If the number of samples after deduplication is insufficient for the sampling budget B, the remaining samples from the candidate pool are supplemented according to the comprehensive score, and the minority class samples are preferentially backfilled. A lower limit quota and an upper limit quota of the number of samples are set for each category, and the quota is linearly distributed according to the category sensitive weighting and the sampling budget B of the current batch; Comprehensive scoring: The four indicators of uncertainty, representativeness, diversity and attribute information are processed in the same scale, and the sample comprehensive value score is calculated in combination with the category sensitive weighting and the sample level expected rarity, so as to select high value samples under the premise of ensuring minority class coverage.

[0010] Preferably, the model training in step five adopts an incremental loss function system driven by class attribute cooperation, which includes a class sensitive rarity modulated classification loss, an attribute consistency cooperative distillation loss and a light regularization loss, specifically: Class sensitive rarity modulated classification loss: The comprehensive weight is constructed in combination with the category sensitive weighting and the sample expected rarity, so that the model gradient update direction is consistent with the sampling logic, and the robustness of the model to rare categories is enhanced; Attribute consistency cooperative distillation loss: The attribute network layer ANL and attribute uncertainty are used to selectively distill the attribute level predictions of the old model and the new model, and through attribute probability distribution difference constraint and attribute representation direction constraint, attribute semantic drift is prevented, and the interpretability and stability of the model in the attribute space are maintained; Light regularization loss: The two norms of the current model parameters and the model parameter snapshot at the end of the last window are calculated to suppress the parameter shock.

[0011] Preferably, the specific implementation of AIAL post-explanation in step six includes: calculating three types of attribute interpretability indicators: attribute uncertainty, which is used to measure the prediction uncertainty of the model on the sample at the attribute prediction layer, and is realized by taking the entropy average of the binary distribution of each attribute; Attribute novelty, which is used to measure the deviation of the sample in the attribute space relative to the prototype of its predicted category, is calculated in the form of Mahalanobis distance; Attribute coverage gap, which is used to measure the coverage and supplement ability of the sample in the attribute dimension to the existing annotation set, is realized by calculating the normalized distance of the sample attribute in the positive un-covered area; Based on the morphological attribute vector extracted by the attribute network layer ANL, the post-explanation of the selected sample at the attribute level is carried out in combination with the knowledge of cell morphology, and the basis for the model to select the sample is explained. The attribute index is combined with the model attention to generate a cell-level feature heat map, intuitively display the key area that the model focuses on, form a readable and verifiable explanation result, and use the explanation information as a feedback signal to assist model diagnosis and data selection for the next round of active learning.

[0012] Preferably, the dynamic mixer in step four schedules the weights of the four types of indicators according to the model training requirements of each stage, increases the weight proportion of the representative and diversity indicators in the Early stage, balances the weight proportions of the four types of indicators in the Middle stage, and increases the weight proportions of the uncertainty and attribute information indicators in the Late stage.

[0013] Preferably, the model fine-tuning in step five adopts a joint training method of high-value samples selected in the current window and historical memory samples, only performs small-step fine-tuning, and monitors the classification accuracy acc, balanced accuracy bal_acc, macro-average F1 value macro-F1 and expected calibration error ECE indicators, and according to the indicator changes, an evaluation signal is fed back to the adaptive mixer to drive the weight update and stage switching of the next window.

[0014] Preferably, the standardization and preprocessing of the unlabeled data in D1 in step two include image denoising, size unification, pixel value normalization operations, and batch division is performed according to a preset fixed number or a dynamically adjusted number standard.

[0015] The beneficial effects of the present application are: The method solves the problems of insufficient response to class imbalance in traditional active learning in cell image analysis and lack of dynamic adaptability of the screening strategy through the synergistic effect of the class imbalance adaptive sample screening mechanism CASS and the dynamic mixer AdaptiveMixer. CASS combines class frequency and effective sample number weighting to significantly improve the screening priority of rare pathological cell samples, cooperates with the three-stage weight scheduling strategy, expands the sample coverage range in the Early stage, balances each type of indicator in the Middle stage, focuses on boundary difficult cases and key attributes in the Late stage, and adds the incremental loss function system driven by class attributes, which not only ensures the recognition ability of the model to the minority class, but also suppresses parameter shock and attribute semantic drift through attribute consistency distillation and light regularization, so that the core indicators such as classification accuracy and balanced accuracy are continuously optimized, and the model generalization and robustness are greatly improved.

[0016] Meanwhile, the method builds a classification prediction + attribute explanation dual-view representation structure through an attribute network layer ANL, combines an AIAL post-hoc explanation method, deeply combines sample high-value judgment basis with cell morphology knowledge, and outputs in a heat map and other visual forms, thereby breaking the black box barrier of traditional active learning, enabling clinical test experts to intuitively verify sample selection logic, and significantly improving the clinical credibility of the technology. Through the continuous iteration of the active learning process, the system realizes a closed loop of data screening, model optimization and explanation feedback, can dynamically adapt to the actual needs of blood cell morphology testing, reduces the cost of manual annotation, and steadily improves the testing accuracy, thereby providing core support for the clinical landing of intelligent cell morphology testing. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a method structure diagram; Figure 2 is a backbone network and attribute network layer construction diagram; Figure 3 is a self-adaptive hybridizer stage switching process schematic diagram; Figure 4 is a sample screening and deduplication process schematic diagram; Figure 5 is an attribute explanation and visualization output flowchart. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0019] It should be noted that all directional and positional indications in the present application, such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "inner", "outer", "top", "low", "lateral", "longitudinal", "center", etc., are only used to explain the relative positional relationship, connection condition, etc. between components in a certain state (as shown in the drawings), and are only for the convenience of describing the present application, and therefore cannot be understood as a requirement for the present application to be constructed and operated in a particular orientation. In addition, the description of "first", "second", etc. in the present application is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features.

[0020] In the description of the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "linking" should be understood in a broad sense, for example, can be fixed connection, can also be detachable connection, or integrally connected; can be mechanical connection; can be directly connected, can also be indirectly connected through intermediate medium, can be internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0021] In the description of the present application, the description of the terms "one embodiment", "some embodiments", "exemplary embodiment", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present application, the exemplary description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0022] S1, in the inspection process, an image is acquired from a blood smear by using a camera to construct a newly collected cell image dataset D1; S2, a small amount of samples are randomly extracted from the newly collected cell image dataset D1 to construct an initial annotation set D2, the D2 is class annotated, and the model is pre-trained based on the D2; the remaining unannotated data in D1 is standardized and preprocessed, and then divided into batches according to the number to form a candidate unannotated pool D3; S3, the system takes D3 as the data flow source of the subsequent process, and inputs the model according to the pre-divided batches. For each batch of input data, the class imbalance adaptive sample screening mechanism CASS is adopted, which comprehensively utilizes the uncertainty, representativeness and diversity of three types of indexes under the current weight configuration, and introduces attribute information, and combines the class frequency and the number of effective samples for weighted calculation, so as to obtain the candidate high-value samples of this batch; S4, in order to make CASS have different emphases in different learning stages, the system schedules the weights of the above four types of indexes through the dynamic mixer (Adaptive Mixer), realizes the gradual switching of Early, Middle and Late three stages; the Early stage emphasizes representativeness and diversity to expand the sample coverage, the Middle stage keeps the weight of four types of indexes relatively balanced, and the Late stage increases the uncertainty and attribute weight to adapt to the fine learning demand of the later model. The weight output by the dynamic mixer will be used to guide the sample screening strategy of CASS in the corresponding stage; S5, after the selection of each batch of samples is completed, the selected new samples are used to fine-tune the model with a small number of steps, and the model is updated by combining the exponential moving average EMA and light regularization; in the evaluation stage, the classification accuracy acc, the balanced accuracy bal_acc, the macro-average F1 value macro-F1 and the expected calibration error ECE are monitored to ensure the stable performance of the model in terms of class balance and calibration; S6, in the active learning process, the attribute information of each batch of selected sample set w is explained after the event using the AIAL method, the reason why the sample is determined to be high value is explained in combination with the knowledge of cell morphology, and the explanation result is visualized and output to enhance the trust of clinical test experts on the system; S7, the system optimized and explained by the above steps is used to enter a new round of blood cell morphology examination, and steps S1-S6 are repeatedly executed to realize the continuous evolution and dynamic optimization of the system.

[0023] The specific method is as follows: The CAS-AL method; Initialization stage: based on the existing cell category data, a newly collected cell data set D1 is constructed. A small amount of samples randomly divided in D1 form an initial annotation set D2, and D2 is annotated for class to be used for model pre-training. The remaining data is input into the system for standardization and preprocessing, and is divided into batches to obtain a candidate unannotated pool D3. The system completes the model structure building and core module initialization in this stage to provide a basis for subsequent active learning.

[0024] Construction of backbone network and attribute network layer; The network includes a low-dimensional feature extractor L, which is composed of the first several residual blocks of a pre-trained model (ResNeXt50), for extracting and outputting basic morphological features ; it also includes a high-dimensional feature extractor H, which is composed of the last several residual blocks of a pre-trained model, for learning and outputting higher-level distinguishing features of cells . Before the class classifier C, an attribute network layer (Attribute Network Layer, ANL) is introduced, which projects the high-dimensional features to the attribute space to obtain the attribute vector: ; The attribute dimensions of cell morphology (such as cell size, nucleus-cytoplasm ratio, particle density, etc.).

[0025] The ANL is composed of two linear mappings , which are responsible for The high-dimensional representation in the model is compressed and aligned to the attribute semantic space. By introducing ANL, the system obtains interpretable attribute representations in addition to the classification output, thus realizing a dual-view representation structure of classification prediction and attribute interpretation.

[0026] (2) Initialization of the strategy unit and mixer; The system instantiates the modules required for the four metrics and builds an adaptive mixer, including: Uncertainty Strategist (U): Employs Entropy to measure the ambiguity of a sample's location on the decision boundary; Representative strategy (R): KNN-density (calculate the top-k cosine similarity mean and normalize to [0,1]) is used to measure the coverage of the overall distribution by the sample; Diversity Strategist (D): Employs the K-Center strategy to encourage the dispersed selection of feature space; Attribute Value Channel (A): Attribute Uncertainty Attribute novelty With attribute coverage gap It is composed of three weighted components (see “Phase 1: Indicator Calculation” for details).

[0027] During the initialization phase, the three-stage weights are provided to the mixer according to a predefined set; the adaptive switching mechanism (see Phase 3) then determines the weights based on batch-to-batch performance consistency metrics. Automatic decision-making. Once active learning begins iterative sampling, the mixer loads the corresponding strategy configuration based on the current stage and dynamically switches between Early, Middle, and Late.

[0028] Phase 1: Samples are taken from the candidate unlabeled pool D3 and input into the backbone network and attribute network layer. Morphological features are extracted through the low-dimensional feature extractor L and the high-dimensional feature extractor H. Four types of indicators, namely uncertainty, representativeness, diversity and attribute value, are calculated to achieve multi-dimensional evaluation of each sample, providing data support for subsequent sample value judgment and class imbalance adaptive screening.

[0029] Indicator Calculation: Uncertainty : Used to measure the model's prediction uncertainty for sample x, employing information entropy: ; in K represents the predicted probability of the model for category c, and K represents the total number of categories.

[0030] Representative The mean cosine similarity of neighborhood density is used to measure the extent to which a sample covers the overall distribution; the larger the value, the more common the region to which the sample belongs.

[0031] ; in The feature vector output by the feature extraction network. These are the k nearest neighbor samples to x.

[0032] diversity The K-Center strategy encourages the selection of samples that are far from the current center set. ; in This is the current set of selected center samples. The feature vector output by the feature extraction network; Attribute Value Used to measure the information value provided by a sample x in the morphological attribute space, due to the uncertainty of its attributes. Attribute novelty With attribute coverage gap It is composed of three weighted combinations: ; in These are the weighting coefficients. , , See Stage Six for the definition.

[0033] Phase Two: Based on the calculation of multiple indicators, a weighted mechanism of class frequency and effective sample number is introduced to perform class-sensitive relabeling and quota constraints on candidate samples. By combining uncertainty, representativeness, diversity and attribute value, a dynamic balance is achieved between prioritizing rare cells and sample distribution, ensuring that the model pays full attention to and utilizes minority class samples during the active learning process.

[0034] Category statistics are weighted by the number of valid samples; During the initialization phase, the class frequency of the labeled samples (including old class samples replayed from memory) is counted. ,in K represents the total number of categories. A category sensitivity score is constructed based on category frequency: ; in, For smoothing parameters.

[0035] Subsequently, the category sensitivity scores are normalized and truncated with upper and lower limits to obtain the final category weighted score: ; in, , representing the mean of sensitivity scores for all categories, where γ>0 indicates the reinforcement index for rare classes. This is the cutoff range for the weighting coefficients. Final weights. This will be used to increase the priority of minority groups in subsequent sample screening stages.

[0036] Category-sensitive recalibration; For each candidate sample x, use the predicted distribution Weighted by category Calculate the expected rarity: ; in This represents the total number of categories.

[0037] Category weighting of Phase 2 output Compared with sample-level expected rarity This will serve as the category constraint and sample rarity adjustment factor for the comprehensive scoring in stage four, and will participate in the final sample selection together with the four weights output from stage three.

[0038] Phase 3: The model integrates four indicators—uncertainty, representativeness, diversity, and attribute value—through an adaptive mixer and uses a phase switching threshold ST to achieve a smooth transition between the Early, Middle, and Late phases. This dynamically adjusts the policy weights, allowing the model to balance exploration and utilization at different phases.

[0039] Three-stage switching strategy: Early stage; Primarily designed for early-stage training when the model structure is not yet stable, the goal is to quickly establish a complete understanding of the data distribution. Enhancing representativeness (R) and diversity (D) ensures comprehensive sampling coverage and sufficient inter-class representation. Uncertainty (U) is treated as a secondary factor to avoid premature concentration on noisy or difficult examples. Attribute value (A) is used to select samples with typical morphological features, improving the quality of initial features.

[0040] Middle stage; Once the model has established a preliminary inter-class structure, it enters the Middle stage, which aims to maintain training stability and class balance. The weights of the four metrics tend to be balanced, taking into account both exploration (D, R) and utilization (U, A). Emphasis is placed on class balance to avoid weakening overall generalization ability due to oversampling of certain classes. The Middle stage is the "steady-state region" of the active learning system and has the most significant impact on training quality.

[0041] Late stage; For cases where the model has converged but still requires improved boundary discrimination capabilities, the following measures are taken: Increase the uncertainty U to pay more attention to difficult boundary examples and confused samples. Add attribute value A to highlight key attribute differences and enhance the identification of minority classes and fine-grained morphologies.

[0042] Adaptive switching mechanism: To provide a consistent measure of overall model performance variation across adjacent batches, a stage switching threshold (SwitchThreshold, ST) based on performance consistency is introduced to comprehensively evaluate batch performance. Compared to the previous batch The performance improvement, consistency, and calibration status. The switching threshold is defined as: ; in ; The four indicators are defined as follows; :batch Overall prediction accuracy on the validation set; :batch Category average recall is used to measure class balance. :batch The macro-average F1 score reflects the ability to distinguish between categories. :batch The expected calibration error reflects the degree of matching between model confidence and prediction accuracy.

[0043] The performance variation between adjacent batches is defined as: ; the remaining , , Defined in the same way.

[0044] The phase switching rules are as follows: like The handover threshold for the current stage (e.g., the handover threshold from Early to Middle) is exceeded h consecutive times. Middle→Late switching threshold If so, proceed to the next stage; like Significantly reduced to the fallback threshold If a core performance indicator continues to degrade, it is permissible to revert from a higher stage to a previous stage to avoid frequent jitter caused by noise.

[0045] In summary, the Adaptive Mixer outputs four weights in each batch. This will be directly used as input to the next stage of the CASS-type imbalanced adaptive sample screening mechanism to guide the comprehensive evaluation, ranking and selection of candidate samples, thereby forming a continuous closed loop of "weight scheduling - sample screening".

[0046] Phase Four: Based on the four weights output from Phase Three, the CASS class imbalanced adaptive screening mechanism is used to perform weighted comprehensive scoring on all candidate samples. After determining the weights of each path, a candidate sample pool is constructed and deduplicated, backfilled, and comprehensively scored. By normalizing and weighted summing the sample indicators, the final comprehensive value score is calculated. Under the premise of ensuring coverage of the minority class, the most informative high-value samples are selected to provide high-quality data input for incremental training of the model.

[0047] Joint constraints of class-sensitive deduplication and category quotas; Class-sensitive similarity threshold ; For candidate samples of the same class c, according to their class sensitivity weights Set similarity threshold ; This deduplication strategy is not applied between different categories to avoid accidentally deleting cross-category samples.

[0048] Backfilling: If the number of samples after deduplication is less than budget B, supplement them from the remaining samples in the candidate pool according to the comprehensive score, and prioritize backfilling of minority classes.

[0049] Category quota constraint range; To prevent majority class samples from consuming the entire budget while ensuring that minority class samples can be included in training, the system sets a minimum quota for the number of selectable samples for each class c. and upper limit quota .

[0050] Quotas are weighted by category. With the sampling budget for this batch Linear allocation: ; in Control the upper and lower limits of intensity. .

[0051] Overall score; For each metric, perform homoscalation (min-max) and combine it with the category-sensitive weighting of the stage two output. Compared with sample-level expected rarity The overall score is: ; ; in Four weights from the output of the adaptive mixer; in The aforementioned expected rarity is obtained by normalizing its value to the [0,1] interval. , used for weighting. , is a hyperparameter for controlling the intensity of the effects of class sensitivity and sample rarity.

[0052] Phase 5: The selected high-value samples and historical memory samples are combined for micro-fine-tune training. By combining EMA smoothing, lightweight regularization and attribute-aware loss function system, the model can be stably updated and knowledge is preserved. At the same time, during the validation phase, metrics such as acc, bal_acc, macro-F1 and ECE are monitored to evaluate model performance and calibration stability.

[0053] Model training and loss function design; This invention constructs a class-attribute collaboratively driven incremental loss function system. Within each window, this system only utilizes "high-value samples selected in the current window" and "historical memory samples" for small-step fine-tuning. It modulates the loss term through class-sensitive weights, expected sample rarity, and attribute interpretation weights, thereby highlighting the contributions of rare classes and key attributes while maintaining model stability.

[0054] The loss function during training consists of three parts: class-sensitive sparse modulation classification loss, attribute consistency co-distillation loss, and lightweight regularization loss. ; Class-Sensitive Rarity Modulation Classification Loss : To ensure that the training process is consistent with the class sensitivity evaluation and sample expected rarity in Phase 2, this invention constructs a class sensitivity rarity modulated classification loss.

[0055] Let the true label of the i-th sample be... The model predicts the probability of this category as: The category sensitivity weights obtained in Phase 2 are: The expected rarity of the sample is The overall weight is defined as follows: ; in is the total number of categories, and c represents the category index. Controlling the intensity of rarefaction modulation The classification loss is then: ; in As a focusing factor, This is the current number of fine-tuned samples.

[0056] This loss ensures that the gradient update direction of the model is completely consistent with the sampling logic of the preceding stage, thereby maintaining robustness to rare classes during continuous active learning.

[0057] Attribute Consistency Collaborative Distillation Loss ; To maintain the semantic stability of the attribute space and enhance the consistency between the old and new models in the attribute space, this invention utilizes the Attribute Network Layer (ANL) constructed in stages one through four and the attribute uncertainty defined in stage six. Selective distillation is performed on the attribute-level predictions of the old and new models.

[0058] Assume the old model in the sample The output attribute logit vector is The new model output is After smoothing with temperature parameter T, the attribute probability vector is obtained: ; Considering the differences in stability among different samples in the attribute space, this invention introduces adaptive weighting coefficients based on attribute uncertainty. This is used to impose stronger consistency constraints on samples with unstable attributes. Its definition is as follows: ; in, The uncertainty of the attributes of sample x. , These represent the mean and standard deviation of the attribute uncertainty within the current batch, respectively. To prevent division by zero of extremely small constants, This is the Sigmoid mapping function.

[0059] Therefore, the attribute consistency collaborative distillation loss is defined as: ; in, , The first term in the equation constrains the confidence of the new model at the attribute level through the difference in attribute probability distribution, while the second term constrains the direction of attribute representation through cosine similarity. This together prevents attribute semantic drift and maintains the interpretability and stability of the model in the attribute space.

[0060] ③ Lightweight regularization loss

[0061] To further suppress drastic parameter oscillations, this invention introduces a lightweight regularization term in addition to the two losses mentioned above: ; in, These are the current model parameters. This is a snapshot of the model parameters at the end of the previous window.

[0062] By minimizing the total loss L during the fine-tuning process of each window, the system achieves the combined effect of class balance constraints, attribute consistency, and parameter smoothing between consecutive windows. This enables the model to quickly absorb new knowledge while maintaining its memory of rare pathological cells and key morphological attributes when faced with a continuously growing stream of blood cell data, thus providing a more stable foundation for subsequent CASS sampling and AIAL interpretation.

[0063] (2) Evaluation and signal reinjection; In each window After fine-tuning, the system calculates four performance metrics for the window on the validation set: { , , , These respectively reflect overall accuracy, category balance, classification discrimination ability, and model calibration degree; Then, these indicators are compared with the previous window. By comparing the corresponding metrics, a performance increment vector is obtained. ; And combined with the change in calibration error Together they constitute the evaluation signal.

[0064] The evaluation signal is fed back to AdaptiveMixer, driving the next window according to the stage switching strategy. The weight updates and phase switching form a learning loop.

[0065] Phase Six: After active learning is completed, the system performs attribute-driven interpretability analysis (AIAL) on the selected sample set in each round. By calculating interpretability indicators at the attribute level and generating a visual heatmap, the system reveals the decision-making basis of the model and feeds the interpretation results back to the system to support the continuous optimization of subsequent active learning.

[0066] (1) Calculation of attribute interpretability index Three types of interpretability indicators were calculated for each selected sample to reveal its selection value: Attribute uncertainty (notation) ); Used to measure the uncertainty of a model's predictions of samples at the attribute prediction head.

[0067] Assume the model outputs the attribute prediction probability vector as follows: Where A is the total number of attributes, and , .

[0068] The uncertainty of an attribute is defined as entropy: ; By taking the entropy average of the binary distributions of each attribute, the model's ambiguity regarding the samples at the attribute level is comprehensively measured. The higher the uncertainty, the greater the potential for information gain from the samples.

[0069] Attribute novelty (denoted as...) ); It is used to measure how much a sample deviates from its predicted class prototype in the attribute space.

[0070] The attribute vector of the sample is The attribute prototype vector of its category is The covariance matrix is .

[0071] Attribute novelty is defined as the Mahalanobis distance form: .; It reflects the "outlierness" of a sample in terms of attribute distribution. When A larger value indicates that the sample differs significantly from the existing training samples in terms of morphological features, and has potential novel information.

[0072] ③ Attribute Coverage Gap (denoted as...) ) It is used to measure the ability of a sample to cover and supplement the existing labeled set in terms of attribute dimensions.

[0073] Let the mean and standard deviation of the current annotation set on the a-th attribute dimension be respectively and The corresponding attribute value of the sample is .

[0074] The coverage gap is defined as the normalized distance of a sample attribute in the "positive uncovered area": ; in This is an adjustable hyperparameter (usually set to 1.0). To prevent division by zero of small constants.

[0075] If a sample exceeds the upper bound of the labeled distribution in some attribute dimensions, it is considered an extension of the knowledge coverage. The larger the value, the more significantly the sample can expand the cognitive space of the model at the attribute level.

[0076] (2) Attribute explanation and visualization output; After active learning is completed, the system uses the morphological attribute vectors extracted by the attribute network layer ANL to provide post-hoc interpretation of the selected samples at the attribute level, helping users understand the basis for the model's selection of these samples (such as karyotype, karyocytic ratio, staining degree, granularity, etc.).

[0077] Simultaneously, by combining attribute metrics and model attention, a cell-level feature heatmap is generated, which intuitively displays the key areas of focus of the model, forming readable and verifiable interpretation results.

[0078] The aforementioned explanatory information will serve as feedback signals for the system, assisting in model diagnosis and data selection for the next round of active learning, thereby enhancing the interpretability and continuously optimizing the system.

[0079] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. An interpretable active learning method driven by the collaborative interaction of cell image class attributes, characterized in that: Includes the following steps: Step 1: During the examination, images are acquired from blood cell smears using a camera to construct a newly acquired cell image dataset D1; Step 2: Randomly select a small number of samples from dataset D1 to construct an initial labeled set D2, label D2 with categories and complete model pre-training based on it; After standardization and preprocessing, the remaining unlabeled data in D1 are divided into several batches according to quantity to form a candidate unlabeled pool D3. Step 3: Input the data in D3 into the model in the preset batches. For each batch of input data, the Class Imbalanced Adaptive Sample Screening Mechanism (CASS) is used. Under the current weight configuration, the three indicators of uncertainty, representativeness and diversity are comprehensively used, and attribute information is introduced. The weighted calculation is performed by combining the class frequency and the number of effective samples to screen out the candidate high-value samples for that batch. Step 4: The AdaptiveMixer is used to schedule the weights of four indicators—uncertainty, representativeness, diversity, and attribute information—to achieve a gradual switching between the Early, Middle, and Late stages. The Early stage emphasizes representativeness and diversity to broaden sample coverage; the Middle stage maintains a relatively balanced weighting of the four indicators; and the Late stage increases the weights of uncertainty and attribute information. The weights output by the AdaptiveMixer are used to guide the sample selection strategy for the corresponding stage. Step 5: After each batch of samples is selected, the model is fine-tuned using the selected new samples in small steps, and the model is updated by combining exponential moving average (EMA) and lightweight regularization; during the evaluation phase, the classification accuracy (acc), balance accuracy (bal_acc), macro-F1 score (macro-F1), and expected calibration error (ECE) are monitored to ensure the model’s stable performance in class balance and calibration. Step Six: During the active learning process, the attribute information of each batch of selected sample sets is interpreted post-hocly using the AIAL method. Combined with cell morphology knowledge, the reasons why the samples are judged to be of high value are explained, and the interpretation results are visualized and output. Step 7: Use the system optimized and interpreted in the above steps to enter a new round of blood cell morphology examination, repeating steps 1-6 to achieve continuous evolution and dynamic optimization of the system.

2. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 1, characterized in that: In step two, before model pre-training, the backbone network and attribute network layers need to be constructed. Specifically, the backbone network includes a low-dimensional feature extractor L and a high-dimensional feature extractor H. The low-dimensional feature extractor L is composed of the first few residual blocks of the pre-trained model ResNeXt50, which is used to extract and output basic morphological features. The high-dimensional feature extractor H consists of several residual blocks of the pre-trained model ResNeXt50, and is used to learn and output higher-level discriminative features of cells. An attribute network layer ANL is introduced before the category classifier C to project the high-dimensional features output by the high-dimensional feature extractor H onto the attribute space to obtain an attribute vector. The cell morphology attribute dimensions corresponding to this attribute vector include cell size, nuclear-cytoplasmic ratio, and particle density. The Attribute Network layer (ANL) consists of two linear mapping layers. It is responsible for compressing and aligning the high-dimensional representation in the high-dimensional features to the attribute semantic space, so that the system can obtain interpretable attribute representations in addition to the classification output, forming a dual-view representation structure of classification prediction and attribute interpretation.

3. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 2, characterized in that: Before model pre-training, the policyr and mixer need to be initialized, which includes: The modules required to instantiate the four-way indicator include the uncertainty strategist U, the representativeness strategist R, the diversity strategist D, and the attribute value channel A; The uncertainty policymaker U uses entropy to measure the degree of ambiguity of a sample at the decision boundary; The representative strategy R adopts KNN-density, which measures the degree to which the sample covers the overall distribution by calculating the Top-k cosine similarity mean and normalizing it to [0,1]. The diversity strategist D employs the K-Center strategy to encourage the dispersed selection of feature space; The attribute value channel A is composed of a weighted combination of attribute uncertainty, attribute novelty, and attribute coverage gap. An adaptive mixer is constructed. During the initialization phase, the mixer is provided with three-stage weights according to a preset definition. The adaptive switching mechanism automatically decides the weight configuration based on the batch performance consistency index ST. After active learning and iterative sampling, the mixer loads the corresponding strategy configuration according to the current stage and dynamically switches between Early-Middle-Late.

4. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 3, characterized in that: The implementation method of the adaptive switching mechanism in step four is as follows: The performance consistency index ST is used to comprehensively measure the performance improvement, consistency and calibration status of the current batch relative to the previous batch. Its calculation is based on the batch’s overall prediction accuracy, class average recall, macro average F1 score and the change in expected calibration error on the validation set. The phase switching rule is as follows: if the performance consistency index ST exceeds the switching threshold of the current phase for h consecutive times, then proceed to the next phase; If the performance consistency index ST drops significantly below the rollback threshold, or if a core performance index continues to degrade, then the system will roll back from the higher stage to the previous stage. The adaptive mixer outputs the weights of four indicators in each batch, which are directly used as input to the Class Imbalanced Adaptive Sample Selection Mechanism (CASS) for the corresponding stage, guiding the comprehensive evaluation, ranking, and selection of candidate samples.

5. The interpretable active learning method driven by collaborative cell image class attributes according to claim 4, characterized in that: The specific implementation process of the Class Imbalanced Adaptive Sample Selection Mechanism (CASS) in step three includes: Category statistics are weighted by the number of valid samples; The class frequencies of labeled samples are statistically analyzed. A class sensitivity score is constructed based on these frequencies. This score is then normalized and truncated to obtain a final class weighting. This weighting is used to prioritize the screening of minority class samples. Class sensitivity relabeling: For each candidate sample, the expected rarity is calculated using the predicted distribution and class weighting, and this rarity is used as a sample rarity adjustment factor in the final sample selection; class-sensitive deduplication and class quota are jointly constrained. For candidate samples of the same category, a deduplication strategy is executed based on a similarity threshold set according to category sensitivity weighting. This deduplication strategy is not executed between different categories. If the number of deduplicated samples is insufficient to meet the selection budget B, samples from the remaining candidate pool are supplemented according to the comprehensive score, with priority given to filling samples from minority classes. Set a lower and upper limit quota for the number of samples for each category. The quota is linearly allocated based on category sensitivity weighting and the sampling budget B for this batch. Comprehensive scoring: The four indicators of uncertainty, representativeness, diversity and attribute information are scaled in the same way. The comprehensive value score of the sample is calculated by combining the category-sensitive weighting and the sample-level expected rarity. High-value samples are selected while ensuring coverage of a minority class.

6. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 5, characterized in that: In step five, the model training employs a class-attribute collaboratively driven incremental loss function system. This loss function system comprises three parts: class-sensitive sparseness modulation classification loss, attribute consistency collaborative distillation loss, and lightweight regularization loss, specifically: Class-sensitive rarity modulation classification loss: By combining category-sensitive weighting with the expected rarity of samples to construct a comprehensive weight, the gradient update direction of the model is kept consistent with the sampling logic, thereby enhancing the robustness of the model to rare categories. Attribute Consistency Collaborative Distillation Loss: Utilizing the Attribute Network Layer (ANL) and attribute uncertainty, selective distillation is performed on the predictions of the old and new models at the attribute level. By constraining the differences in attribute probability distributions and the direction of attribute representations, attribute semantic drift is prevented, thus maintaining the interpretability and stability of the model in the attribute space. Lightweight regularization loss: Suppresses sharp parameter oscillations by calculating the L2 norm of the current model parameters and the snapshot of the model parameters at the end of the previous window.

7. The interpretable active learning method driven by collaborative cell image class attributes according to claim 6, characterized in that: The specific implementation of AIAL post-hoc interpretation in step six includes: calculating three types of attribute interpretability indices: attribute uncertainty, which measures the degree of uncertainty in the model's prediction of samples at the attribute prediction layer, and is achieved by taking the entropy average of the binary distribution of each attribute; Attribute novelty measures how much a sample deviates from its predicted class prototype in the attribute space, and is calculated using Mahalanobis distance. Attribute coverage gap is used to measure the ability of a sample to cover and supplement the existing labeled set in terms of attribute dimension. It is achieved by calculating the normalized distance of the sample attribute in the positive uncovered area. Based on the morphological attribute vectors extracted by the attribute network layer ANL, and combined with cell morphology knowledge, the selected samples are given a post-hoc interpretation at the attribute level, which explains the basis for the model's selection of samples. By combining attribute metrics and model attention, a cell-level feature heatmap is generated, which intuitively displays the key areas that the model focuses on, forming readable and verifiable interpretation results. The interpretation information is used as a feedback signal to assist model diagnosis and data selection for the next round of active learning.

8. The interpretable active learning method driven by the collaborative learning of cell image class attributes according to claim 1, characterized in that: In step four, the dynamic mixer dynamically adjusts the weights of the four types of indicators according to the model training requirements at each stage. In the Early stage, the weight ratio of representativeness and diversity indicators is increased. In the Middle stage, the weight ratio of the four types of indicators is made more balanced. In the Late stage, the weight ratio of uncertainty and attribute information indicators is increased.

9. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 1, characterized in that: In step five, model fine-tuning adopts a joint training method using high-value samples selected in the current window and historical memory samples. Only small-step fine-tuning is performed, while monitoring classification accuracy (acc), balanced accuracy (bal_acc), macro-F1 score (macro-F1), and expected calibration error (ECE). Based on the changes in the indicators, an evaluation signal is generated and injected back into the adaptive mixer to drive the weight update and stage switching of the next window.

10. The interpretable active learning method driven by the collaborative interaction of cell image class attributes according to claim 1, characterized in that: Step two involves the standardization and preprocessing of the unlabeled data in D1, including image noise reduction, size unification, and pixel value normalization. Batch division is performed according to a preset fixed quantity or a dynamically adjusted quantity standard.

Images