Removed unecessary interfaces from MLProvider. Infer endpoint can now…

… return DICOM Seg files.

dicoogle/src/main/java/pt/ua/dicoogle/plugins/PluginController.java

@@ -33,8 +33,8 @@
 import pt.ua.dicoogle.sdk.datastructs.SearchResult;
 import pt.ua.dicoogle.sdk.datastructs.dim.DimLevel;
 import pt.ua.dicoogle.sdk.mlprovider.MLDataset;
-import pt.ua.dicoogle.sdk.mlprovider.MLPrediction;
-import pt.ua.dicoogle.sdk.mlprovider.MLPredictionRequest;
+import pt.ua.dicoogle.sdk.mlprovider.MLInference;
+import pt.ua.dicoogle.sdk.mlprovider.MLInferenceRequest;
 import pt.ua.dicoogle.sdk.mlprovider.MLProviderInterface;
 import pt.ua.dicoogle.sdk.settings.ConfigurationHolder;
 import pt.ua.dicoogle.sdk.task.JointQueryTask;
@@ -837,13 +837,13 @@ public List<Report> indexBlocking(URI path) {
      * @param predictionRequest
      * @return the created task
      */
-    public Task<MLPrediction> makePredictionOverImage(final String provider, final MLPredictionRequest predictionRequest) {
+    public Task<MLInference> infer(final String provider, final MLInferenceRequest predictionRequest) {
         MLProviderInterface providerInterface = this.getMachineLearningProviderByName(provider, true);
         if(providerInterface == null)
             return null;
 
         String taskName = "MLPredictionTask" + UUID.randomUUID();
-        Task<MLPrediction> result = providerInterface.makePrediction(predictionRequest);
+        Task<MLInference> result = providerInterface.infer(predictionRequest);
         result.setName(taskName);
         return result;
     }

dicoogle/src/main/java/pt/ua/dicoogle/server/web/servlets/mlprovider/MakePredictionServlet.java

@@ -3,7 +3,9 @@
 import com.fasterxml.jackson.core.type.TypeReference;
 import com.fasterxml.jackson.databind.JsonNode;
 import com.fasterxml.jackson.databind.ObjectMapper;
+import org.apache.commons.io.FileUtils;
 import org.apache.commons.io.IOUtils;
+import org.apache.commons.lang3.StringUtils;
 import org.dcm4che3.imageio.plugins.dcm.DicomMetaData;
 import org.restlet.data.Status;
 import org.slf4j.Logger;
@@ -13,20 +15,23 @@
 import pt.ua.dicoogle.sdk.datastructs.dim.DimLevel;
 import pt.ua.dicoogle.sdk.datastructs.dim.Point2D;
 import pt.ua.dicoogle.server.web.dicom.WSISopDescriptor;
-import pt.ua.dicoogle.sdk.mlprovider.MLPrediction;
-import pt.ua.dicoogle.sdk.mlprovider.MLPredictionRequest;
+import pt.ua.dicoogle.sdk.mlprovider.MLInference;
+import pt.ua.dicoogle.sdk.mlprovider.MLInferenceRequest;
 import pt.ua.dicoogle.sdk.task.Task;
 import pt.ua.dicoogle.server.web.dicom.ROIExtractor;
 import pt.ua.dicoogle.server.web.utils.cache.WSICache;
 
 import javax.servlet.ServletException;
+import javax.servlet.ServletOutputStream;
 import javax.servlet.http.HttpServlet;
 import javax.servlet.http.HttpServletRequest;
 import javax.servlet.http.HttpServletResponse;
 import java.awt.image.BufferedImage;
+import java.io.File;
 import java.io.IOException;
 import java.io.PrintWriter;
 import java.util.List;
+import java.util.UUID;
 import java.util.concurrent.ExecutionException;
 
 public class MakePredictionServlet extends HttpServlet {
@@ -78,11 +83,11 @@ protected void doPost(HttpServletRequest request, HttpServletResponse response)
 
         String provider = body.get("provider").asText();
         String modelID = body.get("modelID").asText();
-        boolean wsi = body.get("wsi").asBoolean();
+        boolean wsi = body.has("wsi") && body.get("wsi").asBoolean();
         DimLevel level = DimLevel.valueOf(body.get("level").asText().toUpperCase());
         String dimUID = body.get("uid").asText();
 
-        Task<MLPrediction> task;
+        Task<MLInference> task;
 
         if(wsi){
 
@@ -114,11 +119,11 @@ protected void doPost(HttpServletRequest request, HttpServletResponse response)
         task.run();
     }
 
-    private Task<MLPrediction> sendWSIRequest(String provider, String modelID, String baseSopInstanceUID, String uid,
-                                              BulkAnnotation.AnnotationType roiType, List<Point2D> roi, HttpServletResponse response){
+    private Task<MLInference> sendWSIRequest(String provider, String modelID, String baseSopInstanceUID, String uid,
+                                             BulkAnnotation.AnnotationType roiType, List<Point2D> roi, HttpServletResponse response){
 
         ObjectMapper mapper = new ObjectMapper();
-        MLPredictionRequest predictionRequest = new MLPredictionRequest(true, DimLevel.INSTANCE, uid, modelID);
+        MLInferenceRequest predictionRequest = new MLInferenceRequest(true, DimLevel.INSTANCE, uid, modelID);
         BulkAnnotation annotation = new BulkAnnotation();
         annotation.setPoints(roi);
         annotation.setAnnotationType(roiType);
@@ -127,11 +132,11 @@ private Task<MLPrediction> sendWSIRequest(String provider, String modelID, Strin
             DicomMetaData dicomMetaData = this.getDicomMetadata(uid);
             BufferedImage bi = roiExtractor.extractROI(dicomMetaData, annotation);
             predictionRequest.setRoi(bi);
-            Task<MLPrediction> task = PluginController.getInstance().makePredictionOverImage(provider, predictionRequest);
+            Task<MLInference> task = PluginController.getInstance().infer(provider, predictionRequest);
             if(task != null){
                 task.onCompletion(() -> {
                     try {
-                        MLPrediction prediction = task.get();
+                        MLInference prediction = task.get();
 
                         if(prediction == null){
                             log.error("Provider returned null prediction");
@@ -174,26 +179,64 @@ private Task<MLPrediction> sendWSIRequest(String provider, String modelID, Strin
         }
     }
 
-    private Task<MLPrediction> sendRequest(String provider, String modelID, DimLevel level, String dimUID, HttpServletResponse response){
+    private Task<MLInference> sendRequest(String provider, String modelID, DimLevel level, String dimUID, HttpServletResponse response){
         ObjectMapper mapper = new ObjectMapper();
-        MLPredictionRequest predictionRequest = new MLPredictionRequest(true, level, dimUID, modelID);
-        Task<MLPrediction> task = PluginController.getInstance().makePredictionOverImage(provider, predictionRequest);
+        MLInferenceRequest predictionRequest = new MLInferenceRequest(false, level, dimUID, modelID);
+        Task<MLInference> task = PluginController.getInstance().infer(provider, predictionRequest);
         if(task != null){
             task.onCompletion(() -> {
                 try {
-                    MLPrediction prediction = task.get();
+                    MLInference prediction = task.get();
 
                     if(prediction == null){
                         log.error("Provider returned null prediction");
                         response.sendError(Status.SERVER_ERROR_INTERNAL.getCode(), "Could not make prediction");
                         return;
                     }
 
-                    response.setContentType("application/json");
-                    PrintWriter out = response.getWriter();
-                    mapper.writeValue(out, prediction);
-                    out.close();
-                    out.flush();
+                    if(prediction.getDicomSEG() != null){
+                        // We have a file to send, got to build a multi part response
+                        String boundary = UUID.randomUUID().toString();
+                        response.setContentType("multipart/form-data; boundary=" + boundary);
+
+                        ServletOutputStream out = response.getOutputStream();
+
+                        out.print("--" + boundary);
+                        out.println();
+                        out.print("Content-Disposition: form-data; name=\"params\"");
+                        out.println();
+                        out.print("Content-Type: application/json");
+                        out.println(); out.println();
+                        out.print(mapper.writeValueAsString(prediction));
+                        out.println();
+                        out.print("--" + boundary);
+                        out.println();
+                        out.print("Content-Disposition: form-data; name=\"dicomseg\"; filename=\"dicomseg.dcm\"");
+                        out.println();
+                        out.print("Content-Type: application/octet-stream");
+                        out.println(); out.println();
+
+                        byte[] targetArray = new byte[prediction.getDicomSEG().available()];
+                        prediction.getDicomSEG().read(targetArray);
+                        out.write(targetArray);
+                        out.flush();
+                        out.close();
+                    } else {
+                        response.setContentType("application/json");
+                        PrintWriter out = response.getWriter();
+                        mapper.writeValue(out, prediction);
+                        out.close();
+                        out.flush();
+                    }
+
+                    try{
+                        if(!StringUtils.isBlank(prediction.getResourcesFolder())){
+                            FileUtils.deleteDirectory(new File(prediction.getResourcesFolder()));
+                        }
+                    } catch (IOException e){
+                        log.warn("Could not delete temporary file", e);
+                    }
+
                 } catch (InterruptedException | ExecutionException e) {
                     log.error("Could not make prediction", e);
                     try {
@@ -222,7 +265,7 @@ private DicomMetaData getDicomMetadata(String sop) throws IOException{
      * @param scale to transform coordinates
      * @return the ml prediction with the converted coordinates.
      */
-    private void convertCoordinates(MLPrediction prediction, Point2D tl, double scale){
+    private void convertCoordinates(MLInference prediction, Point2D tl, double scale){
         for(BulkAnnotation ann : prediction.getAnnotations()){
             for(Point2D p : ann.getPoints()){
                 p.setX((p.getX() + tl.getX())/scale);

sdk/src/main/java/pt/ua/dicoogle/sdk/mlprovider/MLPrediction.java → sdk/src/main/java/pt/ua/dicoogle/sdk/mlprovider/MLInference.java

@@ -1,21 +1,30 @@
 package pt.ua.dicoogle.sdk.mlprovider;
 
+import com.fasterxml.jackson.annotation.JsonIgnore;
+import org.dcm4che2.io.DicomInputStream;
 import pt.ua.dicoogle.sdk.datastructs.dim.BulkAnnotation;
 
 import java.util.HashMap;
 import java.util.List;
 
 /**
  * This object maps predictions done by the AI algorithms.
- * It can contain a set of metrics and a list of annotations.
+ * It can contain a set of metrics, annotations and a DICOM SEG file.
  */
-public class MLPrediction {
+public class MLInference {
 
     private HashMap<String, String> metrics;
+
     private String version;
 
     private List<BulkAnnotation> annotations;
 
+    @JsonIgnore
+    private String resourcesFolder;
+
+    @JsonIgnore
+    private DicomInputStream dicomSEG;
+
     public HashMap<String, String> getMetrics() {
         return metrics;
     }
@@ -32,6 +41,22 @@ public void setAnnotations(List<BulkAnnotation> annotations) {
         this.annotations = annotations;
     }
 
+    public String getResourcesFolder() {
+        return resourcesFolder;
+    }
+
+    public void setResourcesFolder(String resourcesFolder) {
+        this.resourcesFolder = resourcesFolder;
+    }
+
+    public DicomInputStream getDicomSEG() {
+        return dicomSEG;
+    }
+
+    public void setDicomSEG(DicomInputStream dicomSEG) {
+        this.dicomSEG = dicomSEG;
+    }
+
     public String getVersion() {
         return version;
     }

sdk/src/main/java/pt/ua/dicoogle/sdk/mlprovider/MLPredictionRequest.java → sdk/src/main/java/pt/ua/dicoogle/sdk/mlprovider/MLInferenceRequest.java

@@ -4,7 +4,7 @@
 
 import java.awt.image.BufferedImage;
 
-public class MLPredictionRequest {
+public class MLInferenceRequest {
 
     private boolean isWsi;
 
@@ -16,7 +16,7 @@ public class MLPredictionRequest {
 
     private String modelID;
 
-    public MLPredictionRequest(boolean isWsi, DimLevel level, String dimID, String modelID) {
+    public MLInferenceRequest(boolean isWsi, DimLevel level, String dimID, String modelID) {
         this.isWsi = isWsi;
         this.level = level;
         this.dimID = dimID;

sdk/src/main/java/pt/ua/dicoogle/sdk/mlprovider/MLProviderInterface.java

@@ -34,12 +34,6 @@ public abstract class MLProviderInterface implements DicooglePlugin {
      */
     public abstract MLModel createModel();
 
-
-    /**
-     * This method deploys a model
-     */
-    public abstract void deployModel();
-
     /**
      * This method lists the models created on this provider.
      */
@@ -63,21 +57,6 @@ public abstract class MLProviderInterface implements DicooglePlugin {
      */
     public abstract boolean stopTraining(String trainingTaskID);
 
-    /**
-     * This method creates a endpoint that exposes a service
-     */
-    public abstract void createEndpoint();
-
-    /**
-     * This method lists all available endpoints
-     */
-    public abstract List<MLEndpoint> listEndpoints();
-
-    /**
-     * This method deletes a endpoint
-     */
-    public abstract void deleteEndpoint();
-
     /**
      * This method deletes a model
      */
@@ -87,14 +66,14 @@ public abstract class MLProviderInterface implements DicooglePlugin {
      * Order a prediction over a single object.
      * The object can be a series instance, a sop instance or a 2D/3D ROI.
      *
-     * @param predictionRequest object that defines this prediction request
+     * @param inferRequest object that defines this inference request
      */
-    public abstract Task<MLPrediction> makePrediction(MLPredictionRequest predictionRequest);
+    public abstract Task<MLInference> infer(MLInferenceRequest inferRequest);
 
     /**
-     * This method makes a bulk prediction using the selected model
+     * This method makes a bulk inference request using the selected model
      */
-    public abstract void makeBulkPrediction();
+    public abstract void batchInfer();
 
     public Set<ML_DATA_TYPE> getAcceptedDataTypes() {
         return acceptedDataTypes;